full_review stringlengths 483 5.34k | latex stringlengths 15.5k 101k | paper_url stringlengths 42 44 | arxiv_url stringlengths 32 32 | help_prompt stringlengths 76 408 |
|---|---|---|---|---|
Review of "Cross-language sentiment analysis of European Twitter messages" -- interesting trends analysis but some more approach comparisons and tables for the data would be good.
The authors present an interesting, important and relevant trend analysis of sentiment across languages in several locales during the Covid... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{caption}
\usepackage{url}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\aclfinalcopy %
\setlength\titlebox{10cm}
\newcommand\... | https://openreview.net/forum?id=VvRbhkiAwR | https://arxiv.org/abs/2008.12172 | Please evaluate the paper based on the provided evaluation, focusing on the approach comparisons, data breakdown, and the potential for extended analyses and future improvements. |
Review on "Cross-language sentiment analysis of European Twitter messages during the COVID-19 pandemic"
The authors carried on a deep learning pipeline to analyze the sentiment of Twitter texts, and propose a complete research. The presentation and language part of this submission is good.
However, the research mainl... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{caption}
\usepackage{url}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\aclfinalcopy %
\setlength\titlebox{10cm}
\newcommand\... | https://openreview.net/forum?id=VvRbhkiAwR | https://arxiv.org/abs/2008.12172 | Please evaluate the paper based on its methodology, novelty, and contribution, providing specific feedback on the data statistics, network structure, comparison with other methods, and potential for time series analysis. |
Review
This is a mostly well-written overview of an exercise to assign a sentiment label to the European-country generated tweets during the period December’19-May’20.
The authors describe how they differentiate and identify the country, how they assign the sentiment level (positive, neutral, negative), how they use ... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{caption}
\usepackage{url}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\aclfinalcopy %
\setlength\titlebox{10cm}
\newcommand\... | https://openreview.net/forum?id=VvRbhkiAwR | https://arxiv.org/abs/2008.12172 | Please evaluate the paper based on the methodology used to assign sentiment labels to European-country generated tweets during the period December'19-May'20, including the accuracy of the system and any limitations or potential improvements. |
Excellent description of a critical COVID-19 dataset, some questions remaining
This manuscript describes an exemplary effort to address COVID-19 by bringing together much of the relevant literature into one corpus, CORD-19, and increasing its accessibility by providing a harmonized and standardized format convenient fo... |
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{hyphens}{url}\usepackage{hyperref} %
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\usepackage{enumitem}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{tabularx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{xspac... | https://openreview.net/forum?id=0gLzHrE_t3z | https://arxiv.org/abs/2004.10706 | Please evaluate the clarity and comprehensiveness of my paper, specifically addressing the purpose of my research, the coverage of the literature, and any potential limitations or gaps in the dataset used. |
Overview of a highly important Covid-19 dataset
This is a paper that describes an important research dataset that has been produced during the Covid-19 epidemic. The CORD-19 collection is used for much research and some challenge evaluations. Even though this paper does not report any research results per se, and the p... |
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{hyphens}{url}\usepackage{hyperref} %
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\usepackage{enumitem}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{tabularx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{xspac... | https://openreview.net/forum?id=0gLzHrE_t3z | https://arxiv.org/abs/2004.10706 | Please evaluate the significance and value of the dataset described in this paper for researchers and clinicians in the field of Covid-19. |
CORD-19 is an excellent resource with an impressive integration work for the research community to fight COVID-19.
The authors present the CORD-19 data set and describe how it has been developed and continues to be developed. The CORD-19 data set is a valuable resource that provides access to the latest literature abou... |
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{hyphens}{url}\usepackage{hyperref} %
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\usepackage{enumitem}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{tabularx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{xspac... | https://openreview.net/forum?id=0gLzHrE_t3z | https://arxiv.org/abs/2004.10706 | Please evaluate the paper based on its description and development of the CORD-19 data set, its impact on the biomedical domain, and its potential to support the development of COVID-19 vaccines. |
nice application to new data set to be made available
This paper explores gender differences in linguistic productions between two groups of Redditors who self-identify as either "male" or "female". It examines a corpus of covid-19 pandemic threads in relation to two areas: emotion analysis (employing a VAD lexicon and... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{latexsym}
\usepackage{times}
\usepackage{subcaption}
\usepackage{graphicx}
\usepackage{comment}
\usepackage{color}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{dblfloatfix}
\usepackage{pbox}
\usepackage{array... | https://openreview.net/forum?id=mlmwkAdIeK | https://arxiv.org/abs/2008.05713 | Please evaluate the paper based on its exploration of gender differences in linguistic productions in relation to emotion analysis and topic analysis in a corpus of covid-19 pandemic threads. |
Overall the paper is okay but fails to provide the significance of the work.
This paper aims to understand the difference between male and female discourse in social media looking at a manually annotated set of Reddit threads related to Covid-19 compared to a baseline set. They confirm existing results about male and f... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{latexsym}
\usepackage{times}
\usepackage{subcaption}
\usepackage{graphicx}
\usepackage{comment}
\usepackage{color}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{dblfloatfix}
\usepackage{pbox}
\usepackage{array... | https://openreview.net/forum?id=mlmwkAdIeK | https://arxiv.org/abs/2008.05713 | Please evaluate the significance and novelty of the paper, as well as the clarity of the results and their implications for researchers and practitioners in the field. |
Overall the paper is well written, contains re-usable data, and describes clear results.
Quality:
Overall the paper is well written, contains re-usable data, and describes clear results.
Clarity:
Authors aims of analysis were clearly stated, as were the methods employed. Results are elucidated clearly. The paper is we... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{latexsym}
\usepackage{times}
\usepackage{subcaption}
\usepackage{graphicx}
\usepackage{comment}
\usepackage{color}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{dblfloatfix}
\usepackage{pbox}
\usepackage{array... | https://openreview.net/forum?id=mlmwkAdIeK | https://arxiv.org/abs/2008.05713 | Please evaluate the overall quality, clarity, originality, and significance of my paper. |
timely contribution, could be better positioned with regard to previous work
This paper presents a dataset of 1700 questions related to covid which
are hand labeled and divided into 15 general categories and 207
clusters of semantic equivalence. The value is potentially useful for
general question classification, for s... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{url}
\usepackage{placeins}
\usepackage{nicefrac}
\usepackage{latexsym}
\usepackage{multirow}
\usepackage{float}
\usepackage{booktabs}
\usepackage{graphicx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepacka... | https://openreview.net/forum?id=qd51R0JNLl | https://arxiv.org/abs/2005.12522 | Please evaluate the paper based on its dataset of hand-labeled questions related to COVID-19, considering its potential usefulness for question classification, semantic similarity, and reducing redundancy in question answering. |
A good dataset paper - some clarification and statement of limitation
The paper presents a dataset of questions related to COVID-19; included are annotations mapping questions to specific categories and classes.
It's always good to see papers that contribute datasets to the research community, particularly if they req... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{url}
\usepackage{placeins}
\usepackage{nicefrac}
\usepackage{latexsym}
\usepackage{multirow}
\usepackage{float}
\usepackage{booktabs}
\usepackage{graphicx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepacka... | https://openreview.net/forum?id=qd51R0JNLl | https://arxiv.org/abs/2005.12522 | Please evaluate the clarity and explicitness of the contributions and potential uses of the dataset presented in the paper, as well as the methodology of annotation and any limitations addressed. |
Preliminary work, limited significance
This paper presents a corpus of 1690 documents about COVID-19 manually annotated with questions classes, along with basic question classification algorithms.
Pros
- in scope for the conference
- dataset is publicly available
- technical details are kept to a minimum
- accessible ... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{url}
\usepackage{placeins}
\usepackage{nicefrac}
\usepackage{latexsym}
\usepackage{multirow}
\usepackage{float}
\usepackage{booktabs}
\usepackage{graphicx}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepacka... | https://openreview.net/forum?id=qd51R0JNLl | https://arxiv.org/abs/2005.12522 | Please evaluate the significance and contribution of this paper, as well as the performance of the question classification algorithms and the potential usefulness of the dataset for training QA systems. |
Valuable resource
The paper describes a novel dataset containing answers to a survey that asked participants to describe in short and long text their emotional states and rate their levels of anger, anxiety, desire, disgust, fear, happiness, relaxation, and sadness. The dataset includes responses from 2500 UK resident... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{breaklinks}{hyperref}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{booktabs}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\aclfinalcopy %
\newcommand\BibTeX{B\textsc{ib}\TeX}
\title{Meas... | https://openreview.net/forum?id=JQCYcdHfXyJ | https://arxiv.org/abs/2004.04225 | Please evaluate the paper based on its methodology, dataset, and potential contributions to the NLP community. |
Good first steps toward understanding the UK public's emotional response toward covid19
The focus of this manuscript is to 1) describe a corpus of long and short lengthened emotional responses to the covid19 pandemic by the public, 2) identify meaningful associations between linguistic measures and emotional responses,... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{breaklinks}{hyperref}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{booktabs}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\aclfinalcopy %
\newcommand\BibTeX{B\textsc{ib}\TeX}
\title{Meas... | https://openreview.net/forum?id=JQCYcdHfXyJ | https://arxiv.org/abs/2004.04225 | Please evaluate the paper based on its description of a corpus of emotional responses to the COVID-19 pandemic, the identification of linguistic measures associated with these responses, and the development of prediction models for classifying unseen texts. |
Useful dataset
Summary: The paper introduces a self-reported dataset of 2500 short and long (each) texts about emotions of people during the COVID-19 pandemic. The paper then analyses the dataset in terms of LIWC properties, output of a topic model (most probable topics) and a linear regression model to predict emotion... | \pdfoutput=1
\documentclass[11pt,a4paper]{article}
\PassOptionsToPackage{breaklinks}{hyperref}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{booktabs}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\aclfinalcopy %
\newcommand\BibTeX{B\textsc{ib}\TeX}
\title{Meas... | https://openreview.net/forum?id=JQCYcdHfXyJ | https://arxiv.org/abs/2004.04225 | Please evaluate the usefulness and clarity of my paper, which introduces a self-reported dataset of short and long texts about emotions during the COVID-19 pandemic and analyzes the dataset using LIWC properties, topic modeling, and a linear regression model. |
Interesting correlation, but not actionable and not clear
This paper analyzes the correlation between Covid19 cases (as reported by JHU) and relevant tweets, on a per-country based granularity.
The results are remarkable, showing a very strong correlation (Table 2) for a model trained on Italian data. Moreover, using m... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{times}
\usepackage{latexsym}
\usepackage{times}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{amsmath}... | https://openreview.net/forum?id=ub9_2iAo3D | https://arxiv.org/abs/2006.03202 | Please evaluate the paper based on the usefulness of the correlation between Covid19 cases and relevant tweets, as well as the clarity and reproducibility of the methods used. |
Good idea but some methodological concerns
The following paper hypothesizes that tweets can be used to model COVID-19 outbreaks in across countries. The paper does this by using cross-lingual sentence embeddings from mBERT and LASER to predict case-counts.
Although this is a good idea and Figure 1 is particularly comp... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{times}
\usepackage{latexsym}
\usepackage{times}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{amsmath}... | https://openreview.net/forum?id=ub9_2iAo3D | https://arxiv.org/abs/2006.03202 | Please evaluate the methodology and findings of our paper on using tweets to model COVID-19 outbreaks across countries, including any concerns or suggestions for improvement. |
Review of: Cross-lingual Transfer Learning for COVID-19 Outbreak Alignment
# [REVIEW] Cross-lingual Transfer Learning for COVID-19 Outbreak Alignment
10th June 2020
## SUMMARY
This short paper describes work on cross-lingual transfer learning to track COVID-19 cases across Italy, Thailand, Japan, Turkey, and Indonesi... |
\documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{graphicx}
\usepackage{times}
\usepackage{latexsym}
\usepackage{times}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{amsmath}... | https://openreview.net/forum?id=ub9_2iAo3D | https://arxiv.org/abs/2006.03202 | Evaluate the quality and significance of the paper on "Cross-lingual Transfer Learning for COVID-19 Outbreak Alignment" based on the reviewer's comments. |
Useful resources collected and introduced
This work is an introduction to a suite of resources and their corresponding ideas revolving around the COVID data provided by the Allen institute. Authors in particular provide Neural Covidex which allows exploration of information retrieval and NLP techniques over the COVID d... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{blindtext}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{graphicx}
\usepackage{su... | https://openreview.net/forum?id=PlUA_mgGaPq | https://arxiv.org/abs/2004.05125 | Please evaluate the content and structure of my paper, specifically focusing on the clarity of language, inclusion of evaluation metrics, and overall usefulness for the IR/NLP community. |
Laudable effort, but currently more preprint material
This paper describes the rapid deployment of an IR system for the CORD-19 dataset. The system borrows state-of-the-art pieces, such as T5 and BioBERT, with training on MS MARCO. A search interface is shown that includes potential answer highlighting. Absent hard eva... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{blindtext}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{graphicx}
\usepackage{su... | https://openreview.net/forum?id=PlUA_mgGaPq | https://arxiv.org/abs/2004.05125 | Please evaluate the paper based on its content and presentation, providing feedback on its strengths and weaknesses. |
Evaluation of a rapidly deployed search system for the CORD-19 data set
The authors have quickly stood up a conventional and neural search system based on the CORD-19 dataset, which is a commendable task given the global crisis caused by the Covid-19 pandemic. However, this paper is just a description of their system a... | \documentclass[11pt,a4paper]{article}
\usepackage[hyperref]{acl2020}
\usepackage{times}
\usepackage{latexsym}
\renewcommand{\UrlFont}{\ttfamily\small}
\usepackage{microtype}
\usepackage{amssymb}
\usepackage{amsmath}
\usepackage{blindtext}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{graphicx}
\usepackage{su... | https://openreview.net/forum?id=PlUA_mgGaPq | https://arxiv.org/abs/2004.05125 | Please evaluate the effectiveness and novelty of the search system described in our paper, specifically in comparison to other systems that have been developed for the CORD-19 dataset. |
A novel method of empirical attacks and defenses using semantic transformations
This paper proposed novel defense and attack methods using image transformations. The random transformation defense is designed by averaging the scores of results under random transformations. This paper proposed to use Bayesian optimizatio... |
\documentclass[nohyperref]{article}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{booktabs} %
\usepackage{hyperref}
\newcommand{\theHalgorithm}{\arabic{algorithm}}
\usepackage[accepted]{icml2022}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{mathtools}
\usepackage{amsthm}
\usepackage[capitalize,... | https://openreview.net/forum?id=p4SrFydwO5 | https://arxiv.org/abs/2207.03574 | Please evaluate the proposed defense and attack methods in our paper, which utilize image transformations and Bayesian optimization, and provide your recommendation for acceptance. |
A more reliable evaluation of robustness for random transformation defenses
The authors show that backward-pass differentiable approximation struggles to approximate some transformations accurately, and non-differentiable transformation can’t guarantee robustness. Furthermore, the authors propose a novel attack method ... |
\documentclass[nohyperref]{article}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{booktabs} %
\usepackage{hyperref}
\newcommand{\theHalgorithm}{\arabic{algorithm}}
\usepackage[accepted]{icml2022}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage{mathtools}
\usepackage{amsthm}
\usepackage[capitalize,... | https://openreview.net/forum?id=p4SrFydwO5 | https://arxiv.org/abs/2207.03574 | Please evaluate the robustness evaluation method proposed in our paper and provide feedback on its reliability in assessing the effectiveness of random transformation defenses. |
Convergence Analysis of No-Regret Learning Algorithms in Min-Max Stackelberg Games
This paper provides a convergence proof of no-regret learning algorithms in min-max Stackelberg games. Under certain assumptions, the authors prove that the no-regret learning algorithms will converge to a equilibrium after $T$ iteration... |
\documentclass[sigconf]{aamas}
\usepackage{balance} %
\usepackage{packages}
\usepackage{commands}
\usepackage{mymacros}
\setcopyright{ifaamas}
\acmConference[AAMAS '22]{Proc.\@ of the 21st International Conference
on Autonomous Agents and Multiagent Systems (AAMAS 2022)}{May 9--13, 2022}
{Online}{P.~Faliszewski, V.~M... | https://openreview.net/forum?id=u_lOumlm7mu | https://arxiv.org/abs/2203.14126 | Please evaluate the paper based on its convergence analysis of no-regret learning algorithms in min-max Stackelberg games, including its application to OMD and the study of dynamic Stackelberg games. |
A novel work for Min-Max Stackelberg Games
This work considers min-max Stackelberg games. For two special settings of this problem, this paper proposes no-regret algorithms with convergence guarantee. Moreover, this work provides the theoretical analysis as well as expereimetal results of the algorithms' robustness.
H... |
\documentclass[sigconf]{aamas}
\usepackage{balance} %
\usepackage{packages}
\usepackage{commands}
\usepackage{mymacros}
\setcopyright{ifaamas}
\acmConference[AAMAS '22]{Proc.\@ of the 21st International Conference
on Autonomous Agents and Multiagent Systems (AAMAS 2022)}{May 9--13, 2022}
{Online}{P.~Faliszewski, V.~M... | https://openreview.net/forum?id=u_lOumlm7mu | https://arxiv.org/abs/2203.14126 | Please evaluate the paper based on its proposed algorithms for min-max Stackelberg games, their convergence guarantee, theoretical analysis, and experimental results, while also considering the connection to adversarial machine learning and the formatting of equations and objectives. |
An adversarial detector based on multiple model representations
This paper proposes two approaches using multiple model representations to detect adversarial examples. The authors conducted ablation studies to verify the contribution of the number of underlying models. The major weakness is that comparisons with other ... |
\def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} ... | https://openreview.net/forum?id=LGlhzn1ZJl | https://arxiv.org/abs/2111.07035 | Please evaluate the proposed approaches in our paper that utilize multiple model representations for detecting adversarial examples, considering their effectiveness and any potential weaknesses. |
Interesting idea of ensembling models to detect adversarial instances
This paper proposed to detect adversarial instances by ensembling the deep representations of multiple models in two ways ---- model-wise or unit-wise. The experiments is conducted on CIFAR-10 with the attack methods of FGSM, BIM and CW. The results ... |
\def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} ... | https://openreview.net/forum?id=LGlhzn1ZJl | https://arxiv.org/abs/2111.07035 | Please evaluate the paper based on its proposed method of ensembling deep representations from multiple models to detect adversarial instances, including its strengths and weaknesses. |
Good Approach and Extensive Evaluation
**Summary**:
In this work, the authors have described a novel approach towards detecting adversarial perturbations in images. The approach uses conditional VAEs trained on clean (non-perturbed) images, and leverages that adversarial perturbed examples actually come from a differe... | \documentclass[letterpaper, 10 pt, conference]{IEEEtran}
\IEEEoverridecommandlockouts
\usepackage{cite}
\usepackage{aaai}
\usepackage{subcaption}
\usepackage{amsmath,amssymb,amsfonts}
\usepackage{algorithmicx}
\usepackage[ruled]{algorithm}
\usepackage[noend]{algpseudocode}
\usepackage{graphicx}
\usepackage{textcomp}
... | https://openreview.net/forum?id=Ex1yemaQgU | https://arxiv.org/abs/2111.15518 | Please evaluate the approach and evaluation presented in the paper, focusing on the effectiveness of the proposed method in detecting adversarial perturbations in images. |
Review of Paper 23
This paper propose to use CVAE to detect adversarial examples, while avoid be sensitive on random noisy samples. The experiments are done on MNIST and CIFAR-10, under different attacks and threat models.
The idea of exploiting the information of predicted labels is reasonable, and the authors also e... | \documentclass[letterpaper, 10 pt, conference]{IEEEtran}
\IEEEoverridecommandlockouts
\usepackage{cite}
\usepackage{aaai}
\usepackage{subcaption}
\usepackage{amsmath,amssymb,amsfonts}
\usepackage{algorithmicx}
\usepackage[ruled]{algorithm}
\usepackage[noend]{algpseudocode}
\usepackage{graphicx}
\usepackage{textcomp}
... | https://openreview.net/forum?id=Ex1yemaQgU | https://arxiv.org/abs/2111.15518 | Please evaluate the proposed method of using CVAE to detect adversarial examples, while avoiding sensitivity to random noisy samples, and provide feedback on the experimental results and the potential for over-claiming performance. |
A novel method for diferentiating a bilevel programming solver
This paper proposed a novel method for calculating gradients of a bilevel programming solver. This problem has a wide range of applications like adversarial training, GAN, and combinatorial optimization problems.
Pros
The method is novel and interesting a... |
\documentclass[letterpaper]{article} %
\usepackage{aaai23_arxiv} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\frenchspacing %
\setlength{\pdfpagewidth}{8... | https://openreview.net/forum?id=HvRAM-dpmEv | https://arxiv.org/abs/2302.14473 | Please evaluate the paper based on its proposed method for calculating gradients in a bilevel programming solver, considering its potential applications in areas such as adversarial training, GAN, and combinatorial optimization problems. |
An ensemble training method for robustness
This paper proposes an ensemble training method to enhance the robustness. The ensemble training is implemented by training a super-net called random gated network (RGN). An interesting block called random gated blocks (RGB) is proposed to diversify the vulnerability of differ... | \pdfoutput=1
\documentclass[10pt,twocolumn,letterpaper]{article}
\usepackage{iccv}
\usepackage{times}
\usepackage{epsfig}
\usepackage{graphicx}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage[title]{appendix}
\usepackage{algorithm}
\usepackage{algorithmicx}
\usepackage{algpseudocode}
\usepackage{array}
\algde... | https://openreview.net/forum?id=og7CXiEXqpZ | https://arxiv.org/abs/2103.14795 | Please evaluate the effectiveness and quality of the proposed ensemble training method in enhancing the robustness of your paper. |
An simple yet efficient ensemble training method.
Pros:
1, The proposed ensemble training method is simple yet efficient. Compared with conventional methods, this method is easier to scale up and faster.
2, The paper is easy to follow.
3, Most of the experiments are convincing.
Cons:
1, Though one advantage of the pro... | \pdfoutput=1
\documentclass[10pt,twocolumn,letterpaper]{article}
\usepackage{iccv}
\usepackage{times}
\usepackage{epsfig}
\usepackage{graphicx}
\usepackage{amsmath}
\usepackage{amssymb}
\usepackage[title]{appendix}
\usepackage{algorithm}
\usepackage{algorithmicx}
\usepackage{algpseudocode}
\usepackage{array}
\algde... | https://openreview.net/forum?id=og7CXiEXqpZ | https://arxiv.org/abs/2103.14795 | Please evaluate the proposed ensemble training method in terms of its simplicity, efficiency, scalability, and experimental results. |
Novel method for network verification
The author proposed a polytope traversing algorithm for network verification within a certain region. For Relu networks, the function is piecewise linear and decision regions are partitioned by many polytopes. By traversing these polytopes using a specific algorithm, we can verify ... |
\documentclass[journal]{IEEEtran}
\usepackage[utf8]{inputenc} %
\usepackage[T1]{fontenc} %
\usepackage{hyperref} %
\usepackage{url} %
\usepackage{booktabs} %
\usepackage{amsfonts} %
\usepackage{nicefrac} %
\usepackage{microtype} %
\usepackage{amsmath, amsfonts} %
\usep... | https://openreview.net/forum?id=EQjwT2-Vaba | https://arxiv.org/abs/2111.08922 | Please evaluate the proposed polytope traversing algorithm for network verification in terms of its complexity and scalability, particularly in relation to larger datasets such as ImageNet. |
A novel work for proposing polytope traversing algorithm
This paper points out that ReLU NNs will divide the input domain into many local polytopes. Based on this observation, this work develop a polytope traversing algorithm via BFS and apply in many aspects like local adversarial attacks. The idea is novel and exactl... |
\documentclass[journal]{IEEEtran}
\usepackage[utf8]{inputenc} %
\usepackage[T1]{fontenc} %
\usepackage{hyperref} %
\usepackage{url} %
\usepackage{booktabs} %
\usepackage{amsfonts} %
\usepackage{nicefrac} %
\usepackage{microtype} %
\usepackage{amsmath, amsfonts} %
\usep... | https://openreview.net/forum?id=EQjwT2-Vaba | https://arxiv.org/abs/2111.08922 | Please evaluate the paper based on its proposed polytope traversing algorithm and its potential application in larger tasks such as image classification, considering factors such as computational complexity and the effectiveness of the algorithm in finding optimal solutions. |
An instresting work.
This paper studies robustness indicators of deep models with the goal of better estimating robustness on "unknown" datasets. The authors argue that fixed test sets (e.g., ImageNet-C) are only able to capture a small portion of possible data variations and are limited and prone to generate new over... | \documentclass[10pt,twocolumn,letterpaper]{article}
\usepackage{wacv}
\usepackage{times}
\usepackage{epsfig}
\usepackage{graphicx}
\usepackage{amsmath}
\usepackage{amssymb}
\def\wacvPaperID{****} %
\ifwacvfinal
\def\assignedStartPage{9876} %
\fi
\ifwacvfinal
\usepackage[breaklinks=true,bookmarks=false]{hyperref}
\el... | https://openreview.net/forum?id=UHBsuFPrJ11 | https://arxiv.org/abs/2106.12303 | Please evaluate the paper based on its empirical approach to understanding the robustness of deep models' inner feature space and its focus on estimating robustness on "unknown" datasets. |
A new perspective for assessing robustness from the clustering performance of the model's latent space.
**Pros:**
Generally this paper proposes a new indicator for the model's robustness based on the clustering performance of the model's latent space. Also, vast experiments demonstrate the correlation between the indic... | \documentclass[10pt,twocolumn,letterpaper]{article}
\usepackage{wacv}
\usepackage{times}
\usepackage{epsfig}
\usepackage{graphicx}
\usepackage{amsmath}
\usepackage{amssymb}
\def\wacvPaperID{****} %
\ifwacvfinal
\def\assignedStartPage{9876} %
\fi
\ifwacvfinal
\usepackage[breaklinks=true,bookmarks=false]{hyperref}
\el... | https://openreview.net/forum?id=UHBsuFPrJ11 | https://arxiv.org/abs/2106.12303 | Please evaluate the proposed indicator for assessing the robustness of our model based on the clustering performance of the latent space, and provide your thoughts on its effectiveness compared to existing methods. |
Lacks Novelty
**Summary**:
The paper talks about how adversarial perturbations can affect the performance of OOD detectors. They present a unified framework to study both adversarial attacks on in-distribution as well as OOD inputs (ie attacks which aim to increase both FP and FN). An algorithm called ALOE is present... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=WMIoz7O_DPz | https://arxiv.org/abs/2003.09711 | Please evaluate the novelty and contribution of our paper in the context of adversarial perturbations and their impact on the performance of out-of-distribution (OOD) detectors. |
Review
**Summary of the paper:**
This paper extensively studies the problem of robust OOD detection on common OOD detection approaches and shows that existing OOD detection algorithms can be easily attacked to produce mistaken OOD prediction by adding small perturbations to the in-distribution and OOD inputs. To addre... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=WMIoz7O_DPz | https://arxiv.org/abs/2003.09711 | Please evaluate the effectiveness of the proposed method in improving robust out-of-distribution detection performance compared to existing state-of-the-art methods, based on the evaluation results presented in the paper. |
Review of Paper4
**Summary of the paper:**
The authors propose a novel diversity-promoting learning approach for the deep ensembles to overcome the shared vulnerabilities in its members. Experiments on MNIST, Fashion-MNIST, and CIFAR-10 are conducted to demonstrate the efficacy of the proposed methods.
**Main Review... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=wGkmGrDsco8 | https://arxiv.org/abs/2112.03615 | Please evaluate the effectiveness of the proposed method in improving the adversarial robustness of deep ensembles, considering its simplicity and the results presented in the experiments. |
Saliency Diversified Deep Ensemble for Robustness to Adversaries
This paper proposes a diversity-promoting learning approach for the deep ensembles, which promotes saliency map diversity (SMD) on ensemble members to prevent the attacker from targeting all ensemble members by introducing an additional term. Thus it can ... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=wGkmGrDsco8 | https://arxiv.org/abs/2112.03615 | Please evaluate the proposed approach in the paper for promoting saliency map diversity in deep ensembles to improve robustness to adversaries. |
A new normalization method.
The paper introduces a new normalization method named pixel-wise tensor normalization which improves both accuracy and robustness of the model. However, the results shows somewhat improvement, but not significant. Also, I think the paper is not providing enough theoretical backups for the cl... | \documentclass[10pt, a4paper, onecolumn]{article}
\usepackage[numbers]{natbib}%
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage{natbib}
\usepackage[utf8]{inputenc} %
\usepackage{booktabs} %
\usepackage{amsfonts} %
\usepackage{nicefrac} %
\usepackage{microtype} %
\usepackage{xcolor} ... | https://openreview.net/forum?id=035VtDXUjLN | https://arxiv.org/abs/2109.02345 | Please evaluate the effectiveness and theoretical support of the proposed normalization method in my paper. |
Review of paper "Tensor Normalization and Full Distribution Training"
This paper proposes two techniques including tensor normalization and full distribution training to improve model robustness. These two techniques are easy to understand, and bring improved robustness compared with the baseline. Here are some suggest... | \documentclass[10pt, a4paper, onecolumn]{article}
\usepackage[numbers]{natbib}%
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage{natbib}
\usepackage[utf8]{inputenc} %
\usepackage{booktabs} %
\usepackage{amsfonts} %
\usepackage{nicefrac} %
\usepackage{microtype} %
\usepackage{xcolor} ... | https://openreview.net/forum?id=035VtDXUjLN | https://arxiv.org/abs/2109.02345 | Please evaluate the proposed techniques of tensor normalization and full distribution training in terms of their integration with adversarial training and potential advantages beyond adversarial robustness. |
Meta Adversarial Perturbations
This paper proposes a meta adversarial perturbation (MAP) and obtains a better initialization that causes natural images to be misclassified with high probability, which is only updated through a one-step gradient ascent update.
Experimental performance can be demonstrated that the metho... | \documentclass[english]{article}
\usepackage[T1]{fontenc}
\usepackage[latin9]{inputenc}
\usepackage{array}
\usepackage{float}
\usepackage{multirow}
\usepackage{amstext}
\usepackage{amssymb}
\usepackage{graphicx}
\makeatletter
\providecommand{\tabularnewline}{\\}
\floatstyle{ruled}
\newfloat{algorithm}{tbp}{loa}
\prov... | https://openreview.net/forum?id=gP4WxGjNd3k | https://arxiv.org/abs/2111.10291 | Please evaluate our paper by considering its performance on various datasets, such as ImageNet, and comparing it to other baselines like DIM and TIM. |
A good work for improving the robustness against unauthorized replication attacks
This work proposes Constrained Randomization of Policy (CRoP) as a deep reinforcement learning defence method against unauthorized replication attacks. Though the theoretical part needs to be organized better, the extensive experiments sh... | \documentclass{article}
\usepackage[final, nonatbib]{neurips_2021}
\usepackage[utf8]{inputenc} %
\usepackage[T1]{fontenc} %
\usepackage{hyperref} %
\usepackage{url} %
\usepackage{booktabs} %
\usepackage{amsfonts} %
\usepackage{nicefrac} %
\usepackage{microtype} %
\usepackage... | https://openreview.net/forum?id=o_O7TOBC7jl | https://arxiv.org/abs/2109.14678 | Please evaluate the effectiveness of the proposed Constrained Randomization of Policy (CRoP) method in improving the robustness against unauthorized replication attacks in the paper. |
Review of paper "Is AutoAttack/AutoBench a suitable Benchmark for Adversarial Robustness?"
This paper discusses several limitations of AutoAttack and RobustBench for robustness evaluation. It points out two limitations: 1) The adversarial examples generated by AutoAttack can be easily detected. 2) AutoAttack does not g... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=aLB3FaqoMBs | https://arxiv.org/abs/2112.01601 | Please evaluate the paper "Is AutoAttack/AutoBench a suitable Benchmark for Adversarial Robustness?" based on its findings and experiments. |
An Adversarial Benchmark for Fake News Detection Models
This paper proposes an adversarial benchmark for fake news detection, which is designed to evaluate models’ ability to reason about real-world facts. And some findings can strengthen the need for fake news classification models.
However, the author is also expect... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=n3PMOhS42s6 | https://arxiv.org/abs/2201.00912 | Please evaluate the proposed benchmark for fake news detection in terms of its ability to assess models' reasoning about real-world facts and its potential to strengthen the need for fake news classification models. |
Timely research with exciting results, more details of computational gain would make the work stronger
The paper proposes an operator learning approach to learn a mapping between initial and final states of the droplet coalescence process to enable rapid and accurate part-scale build simulation. Authors compare this ap... | \pdfoutput=1
\def\year{2022}\relax
\documentclass[letterpaper]{article}
\usepackage{aaai22}
\usepackage[hyphens]{url}
\usepackage{graphicx}
\urlstyle{rm}
\def\UrlFont{\rm}
\usepackage{graphicx}
\usepackage{natbib}
\usepackage{caption}
\DeclareCaptionStyle{ruled}%
{labelfont=normalfont,labelsep=colon,strut=off}
\... | https://openreview.net/forum?id=FuojrywNwIM | https://arxiv.org/abs/2202.03665 | Please evaluate the paper based on its demonstration of the operator learning approach for rapid and accurate part-scale build simulation, including the analysis of computational gain and the selection process for training data. |
The paper presents a new Fourier Neural Operator (FNO) approach to accelerate manufacturing simulations
The paper presents a new Fourier Neural Operator (FNO) approach to accelerate metal jet droplet deposition manufacturing simulations. The proposed approach is an order of magnitude faster than the existing state-of-t... | \pdfoutput=1
\def\year{2022}\relax
\documentclass[letterpaper]{article}
\usepackage{aaai22}
\usepackage[hyphens]{url}
\usepackage{graphicx}
\urlstyle{rm}
\def\UrlFont{\rm}
\usepackage{graphicx}
\usepackage{natbib}
\usepackage{caption}
\DeclareCaptionStyle{ruled}%
{labelfont=normalfont,labelsep=colon,strut=off}
\... | https://openreview.net/forum?id=FuojrywNwIM | https://arxiv.org/abs/2202.03665 | Please evaluate the paper based on its proposed Fourier Neural Operator (FNO) approach for accelerating metal jet droplet deposition manufacturing simulations, including its advantages, limitations, and potential areas for further investigation. |
The paper presents a Generative Adversarial Network-based design methodology which allows uncertainty quantification (UQ) of geometric variability. The main idea is to learn a low-D representation of possibly high-D nominal design spaces, and quantify the uncertainty through the learning of a conditional posterior dist... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=rm4rxTrrTjd | https://arxiv.org/abs/2112.08919 | Please evaluate the paper based on its novel use of Generative Adversarial Networks (GANs) to learn a reduced design space and quantify manufacturing uncertainty, and provide insights on the effect of noise in creating the fabricated design space and the selection criteria for initial samples in the Bayesian optimizati... |
A novel approach to Robust Design leveraging conditional generative models
This paper addresses the use of generative models for design under uncertainty, such as in the case where manufacturing variability may affect a design's performance or feasibility. The key idea of the paper is that rather than modeling design u... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=rm4rxTrrTjd | https://arxiv.org/abs/2112.08919 | Please evaluate the paper based on its approach to using generative models for design under uncertainty and its novel idea of modeling uncertainty as a conditional distribution over a given nominal design. |
Well written paper for GP based system ID for optical fiber communication networks
This paper presents a GP approach to create a data-driven emulatorof a detailed physics optical fiber communication network simulator. The GP approach naturally accounts for parameter uncertainty and serves as a robust surrogate model th... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=ug3MANo4x8z | https://arxiv.org/abs/2202.11700 | Please evaluate the paper on GP based system identification for optical fiber communication networks, specifically addressing the clarity of the SSFM approach's computational cost in creating the dataset and the impact of synthetic noise on parameter estimation. |
Review of Gaussian Process-Driven History Matching for Physical Layer Parameter Estimation in Optical Fiber Communication Networks
This paper has demonstrated a GP-driven history matching (HM) approach for parameter estimation/calibration at the physical layer of a optical fiber comm network. The authors have laid out ... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=ug3MANo4x8z | https://arxiv.org/abs/2202.11700 | Please evaluate the paper on Gaussian Process-Driven History Matching for Physical Layer Parameter Estimation in Optical Fiber Communication Networks, focusing on its suitability for parameter estimation, clarity of algorithms, scalability with dimension, and performance in high-dimensional parameter spaces. |
Interesting work
The paper proposes a new Grassmanian manifold based shape representation for airfoils. The authors suggest that such a representation may be more conducive to AI/ML algorithms. They also show the advantages of Grassmanian representations over CST, the current state of art representations for airfoils.
... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=1RRU6ud9YC | https://arxiv.org/abs/2201.04649 | Please evaluate the paper based on its proposed Grassmanian manifold based shape representation for airfoils, its comparison to the current state of the art representation, and its potential advantages for AI/ML algorithms. |
Review for the novel Grassmannian Shape Representations of Aerodynamic Applications
Summary: this paper presents a novel airfoil shape design method based on data-driven techniques. Specifically, the authors proposed a novel representation of shapes which decouples affine-style deformations from a rich set of data-driv... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=1RRU6ud9YC | https://arxiv.org/abs/2201.04649 | Please evaluate the practicality and complexity of applying data-driven techniques for airfoil shape design in the aerospace domain, as presented in our paper. |
The authors have presented a GAN-based cloning strategy that helps in reducing the probability estimation variance for rare events. An optimization problem is solved to control the disturbance induced by cloning on the density function.
The advantages of GAN-based cloning strategies over random cloning have been clearl... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=e6k_JgCT1P | https://arxiv.org/abs/2112.15444 | Please evaluate our paper, specifically addressing the effectiveness of our GAN-based cloning strategy in reducing probability estimation variance for rare events and any potential limitations or suggestions for improvement. |
A GAN based improved on rare event estimation methods.
This paper primarily concerns methods for reducing the variance in importance splitting-based probability estimators in the context of rare events. The paper argues that the standard strategy in Genealogical importance splitting—random splitting—can fail in cases w... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\usepackage{natbib} %
\usepackage{caption} %
\DeclareCaptionStyle{ruled}{lab... | https://openreview.net/forum?id=e6k_JgCT1P | https://arxiv.org/abs/2112.15444 | Please evaluate the paper based on its proposed method of using a GAN to improve rare event estimation methods, including its background, motivation, experimental results, and potential applications. |
Nice new method for developing deep surrogate PDE solvers in a data-efficient manner
The paper proposes a natural approach for improving sample (data) efficiency in deep surrogate PDE solvers. The high level idea is to use two spatial scales (first train a "low-fidelity" model and use its outputs as side information fo... | \documentclass{nature}
\usepackage[ruled,vlined]{algorithm2e}
\usepackage{amssymb}
\usepackage{xcolor}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}
\usepackage{endfloat} %
\date{\today}
\newcommand{\citeasnoun}[1]{Ref.~\citenum{#1}}
\newcommand{\secref}[1]{Sec.~\ref{#1}}
\newcommand{\Secref}[1]{Section~\ref{#1}}... | https://openreview.net/forum?id=vQmS8ueWIFm | https://arxiv.org/abs/2111.05841 | Please evaluate the proposed method for developing deep surrogate PDE solvers in terms of its sample efficiency and potential for extension to a hierarchy of coarse-to-fine mappings. |
Coarse-to-fine physics informed mapping, promising results
This paper presents a method (PEDS) to construct physics informed surrogate model using a coarse to fine framework where a NN is used to map the fine input to a 'generated' coarse input which is then combined with a coarse version of the fine input and fed to a... | \documentclass{nature}
\usepackage[ruled,vlined]{algorithm2e}
\usepackage{amssymb}
\usepackage{xcolor}
\usepackage[utf8]{inputenc}
\usepackage{amsmath}
\usepackage{endfloat} %
\date{\today}
\newcommand{\citeasnoun}[1]{Ref.~\citenum{#1}}
\newcommand{\secref}[1]{Sec.~\ref{#1}}
\newcommand{\Secref}[1]{Section~\ref{#1}}... | https://openreview.net/forum?id=vQmS8ueWIFm | https://arxiv.org/abs/2111.05841 | Please evaluate the paper based on the presented method of constructing a physics informed surrogate model using a coarse to fine framework and its performance compared to other methods. |
Solid paper
The paper proposes a method for anomaly segmentation - a problem widely observed in manufacturing. For that purpose, they leverage hard augmentation, self-supervised learning for generation, and a discriminator for anomaly detection. Anoseg provides promising results when compared to existing methods.
Ove... |
\documentclass{article} %
\usepackage{iclr2022_conference,times}
\usepackage{amsmath,amsfonts,bm}
\newcommand{\figleft}{{\em (Left)}}
\newcommand{\figcenter}{{\em (Center)}}
\newcommand{\figright}{{\em (Right)}}
\newcommand{\figtop}{{\em (Top)}}
\newcommand{\figbottom}{{\em (Bottom)}}
\newcommand{\captiona}{{\em (a... | https://openreview.net/forum?id=x-Tw-P777R | https://arxiv.org/abs/2110.03396 | Please evaluate the clarity and comprehensibility of the paper, including the flow of the proposed method and the inclusion of relevant works in the field of anomaly segmentation using deep generative models. |
Conceptually clear paper with convincing results. Additional details will make paper easier to follow
This paper discusses a method for anomaly segmentation using ideas from multiple fields, including hard augmentation to generate artificial data with anomalies, adversarial training for improved generalization, and coo... |
\documentclass{article} %
\usepackage{iclr2022_conference,times}
\usepackage{amsmath,amsfonts,bm}
\newcommand{\figleft}{{\em (Left)}}
\newcommand{\figcenter}{{\em (Center)}}
\newcommand{\figright}{{\em (Right)}}
\newcommand{\figtop}{{\em (Top)}}
\newcommand{\figbottom}{{\em (Bottom)}}
\newcommand{\captiona}{{\em (a... | https://openreview.net/forum?id=x-Tw-P777R | https://arxiv.org/abs/2110.03396 | Please evaluate the clarity and comprehensiveness of the paper, including the presentation of the concept and the results. |
Differential property prediction
Authors use XGBoost and MLP to predict property differences for different material pairs. While the paper is for most part well written, I am not convinced of the motivation, novelty, and experimental setting. The current problem can be handled using a learning to rank model, for which ... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\pdfoutput=1
\usepackage{amsmath}
\usepackage{amsthm}
\usepackage{amssymb}
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\u... | https://openreview.net/forum?id=SbCndr5Yu6T | https://arxiv.org/abs/2112.01687 | Please evaluate the motivation, novelty, and experimental setting of our paper, specifically addressing the use of XGBoost and MLP for predicting property differences in different material pairs, the potential use of learning to rank models, the number of training/test samples, and how we handle overfitting and the cur... |
Not fully convinced of the motivation of the approach, but well-written paper.
This paper provides the differential property classification framework (DPC), which converts the regression into a classification problem. The author proposes labeling three different classes based on the difference of two input processing p... | \def\year{2022}\relax
\documentclass[letterpaper]{article} %
\pdfoutput=1
\usepackage{amsmath}
\usepackage{amsthm}
\usepackage{amssymb}
\usepackage{aaai22} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\urlstyle{rm} %
\def\UrlFont{\rm} %
\u... | https://openreview.net/forum?id=SbCndr5Yu6T | https://arxiv.org/abs/2112.01687 | Please evaluate the paper based on the justification for converting the regression problem into a classification problem using the differential property classification framework (DPC). |
Meta-review
The paper presents the system description of Team ÚFAL for the CMCL 2022 Shared Task on Multilingual and Crosslingual Prediction of Human Reading Behavior. The authors present a range of model comparisons for eye-tracking prediction. As pointed out by the reviewers, the paper has potential but requires some... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{amsmath} %
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage[capitalize]{cleveref}
\Crefformat{figure}{#2Fig.~#1#3}
\Crefmultiformat{figure}{Figs.~#2#1#3}{ and~#2#1#3}{, #2#1#3}{ and~#2#1#3}
\usepackage{times}
\usepackage{latexsym}
\usepac... | https://openreview.net/forum?id=HF-ez2Bi7-9 | https://arxiv.org/abs/2204.04998 | Please evaluate the paper based on its system description for the CMCL 2022 Shared Task on Multilingual and Crosslingual Prediction of Human Reading Behavior, focusing on the clarity of the model architectures and the analysis of the results. |
High potential, but further work is needed
The paper describes the system proposed for the CMCL2022 Shared Task on Multilingual and Crosslingual Prediction of Human Reading Behavior, and the different architectures implemented and compared by the Team ÚFAL.
**Pros**:
interesting approach, the different results of the ... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{amsmath} %
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage[capitalize]{cleveref}
\Crefformat{figure}{#2Fig.~#1#3}
\Crefmultiformat{figure}{Figs.~#2#1#3}{ and~#2#1#3}{, #2#1#3}{ and~#2#1#3}
\usepackage{times}
\usepackage{latexsym}
\usepac... | https://openreview.net/forum?id=HF-ez2Bi7-9 | https://arxiv.org/abs/2204.04998 | Please evaluate the paper based on the provided information, focusing on the justification for the different system architectures and the depth of analysis in the results section. |
Potentially informative contribution, but missing discussion of results
This paper presents the system description for a contribution to the CMCL 2022 Shared Task on Multilingual and Crosslingual Prediction of Human Reading Behavior. The authors propose to tackle to task using pretrained language models and performed a... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{amsmath} %
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage[capitalize]{cleveref}
\Crefformat{figure}{#2Fig.~#1#3}
\Crefmultiformat{figure}{Figs.~#2#1#3}{ and~#2#1#3}{, #2#1#3}{ and~#2#1#3}
\usepackage{times}
\usepackage{latexsym}
\usepac... | https://openreview.net/forum?id=HF-ez2Bi7-9 | https://arxiv.org/abs/2204.04998 | Please evaluate the strengths and weaknesses of our paper on the system description for the CMCL 2022 Shared Task on Multilingual and Crosslingual Prediction of Human Reading Behavior, focusing on the systematic study of different model configurations and the analysis of contributions to performance. |
Very well described details of the working system
## Summary
This paper describes the system of NU-HLT for the CMCL 2022 shared task. The system, inspired by previous works in speech recognition, uses a novel preprocessing step involving the transformation of words to a global vector space using IPA. Next fourteen fea... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{amssymb}
\usepackage{pdfpages}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{caption}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\title{... | https://openreview.net/forum?id=B0lg2tPwOxc | https://arxiv.org/abs/2202.10855 | Please evaluate the clarity and informativeness of the system description in my paper, as well as the novelty of the idea of transforming raw words into IPA. |
Good descriptions of a working system
### Summary
This paper describes the NU-HLT system in CMCL 2022 shared task. This system is inspired by both classical and recent previous works in speech recognition systems. First, the raw terms are transformed into a global shared space using IPA. Then, some features, including ... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{amssymb}
\usepackage{pdfpages}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{booktabs}
\usepackage{multirow}
\usepackage{caption}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\title{... | https://openreview.net/forum?id=B0lg2tPwOxc | https://arxiv.org/abs/2202.10855 | Please evaluate the paper based on the clarity of the system description, the novelty and effectiveness of the IPA transformation, and the reporting of results and correlations. |
Interesting idea and reasonable experiment results
This paper proposed an interesting idea to learn the Tokenizer/Vocabulary for federated language models. The previous word-level tokenizer/vocabulary can be potentially generated by the following methods: (1) use a public dataset, which may have distribution shift comp... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{subfigure}
\usepackage{booktabs}
\usepackage{threeparttable}
\usepackage{xspace}
\AtBeginDocument{%
\... | https://openreview.net/forum?id=rhz7nqYfF-q | https://arxiv.org/abs/2203.09943 | Please evaluate the proposed method for learning the Tokenizer/Vocabulary for federated language models and its experimental results. |
Review
This paper proposes a federated learning framework to train a tokenizer while it does not require additional privacy budget in differential privacy. Training the tokenizer is an important part of learning a language model (e.g., Transformer and BERT), but to my best knowledge, it is the first work to study how t... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{subfigure}
\usepackage{booktabs}
\usepackage{threeparttable}
\usepackage{xspace}
\AtBeginDocument{%
\... | https://openreview.net/forum?id=rhz7nqYfF-q | https://arxiv.org/abs/2203.09943 | Please evaluate the proposed federated learning framework for training a tokenizer in the context of language models, specifically addressing the system and privacy model, as well as comparing the proposed scheme to directly training the tokenizer based on private dataset in private federated learning. |
Review
This paper provides a novel method on training a tokenizer along with the language model privately in a federated learning setting. By utilizing the post-processing theorem of differential privacy, the authors claim that the proposed method satisfies DP without additional privacy cost on training the tokenizer. ... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{subfigure}
\usepackage{booktabs}
\usepackage{threeparttable}
\usepackage{xspace}
\AtBeginDocument{%
\... | https://openreview.net/forum?id=rhz7nqYfF-q | https://arxiv.org/abs/2203.09943 | Please evaluate the clarity of the privacy protection measures and the utility performance of the proposed method in training a tokenizer along with a language model in a federated learning setting, compared to the heavy-hitters algorithm. |
Reasonable idea, good results, but more work needs to be done before this can be used in practice
Thanks for the submission! I enjoyed reading this paper. The goal of this paper is to improve the tokenizer using the samples matching the real distribution of the data without incurring an additional PFL budget. The basic... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{subfigure}
\usepackage{booktabs}
\usepackage{threeparttable}
\usepackage{xspace}
\AtBeginDocument{%
\... | https://openreview.net/forum?id=rhz7nqYfF-q | https://arxiv.org/abs/2203.09943 | Please evaluate the proposed algorithm in terms of its practical applicability, potential limitations, and the impact it may have on training time and overall performance. |
The authors propose a method to train a 'matched' tokenizer alongside the decentralized and private federated learning of an NLP model over the client data.
The authors propose a method to train a 'matched' tokenizer alongside the decentralized and private federated learning of an NLP model over the client data. In par... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{graphicx}
\usepackage{subfigure}
\usepackage{booktabs}
\usepackage{threeparttable}
\usepackage{xspace}
\AtBeginDocument{%
\... | https://openreview.net/forum?id=rhz7nqYfF-q | https://arxiv.org/abs/2203.09943 | Please evaluate the proposed method for training a 'matched' tokenizer alongside decentralized and private federated learning of an NLP model, specifically in terms of its effectiveness in maintaining accuracy while preserving privacy. |
Adapting intrinsic gradient compression in federated settings
Summary:
A gradient compression technique for federated settings based on the intrinsic dimension concept is proposed. Three variations of the technique are implemented and their tradeoffs in terms of parameter exploration, federation performance and uplink ... | \pdfoutput=1
\def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage[preprint,nonatbib]{neurips_2021} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\usepackage{amsmath}
\usepackage{booktabs}
\urlstyle{rm} %
\def\UrlFont{\rm}... | https://openreview.net/forum?id=H3NUh9Kft-c | https://arxiv.org/abs/2112.02656 | Evaluate the proposed gradient compression technique for federated settings based on the concept of intrinsic dimension, considering its strengths and weaknesses, empirical evaluation against baselines, and potential applications in defending against inference attacks. |
Adopting a classical sketching idea into federated learning applications
This paper proposes to use a classical sketching idea in federated SGD-like algorithms to improve the communication efficiency of federated learning algorithms. The proposed method has been used widely in centralized distributed SGD with good effi... | \pdfoutput=1
\def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage[preprint,nonatbib]{neurips_2021} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\usepackage{amsmath}
\usepackage{booktabs}
\urlstyle{rm} %
\def\UrlFont{\rm}... | https://openreview.net/forum?id=H3NUh9Kft-c | https://arxiv.org/abs/2112.02656 | Please evaluate the proposed method of adopting a classical sketching idea in federated learning algorithms to improve communication efficiency, considering its potential novelty and the possibility of applying the sketching idea on top of federated averaging. Additionally, discuss the potential impact of the error fee... |
The authors propose a set of communication-efficient federated learning algorithms that are based on the prior idea of intrinsic dimension in theoretical machine learning.
The authors propose a set of communication-efficient federated learning algorithms that are based on the prior idea of intrinsic dimension in theore... | \pdfoutput=1
\def\year{2022}\relax
\documentclass[letterpaper]{article} %
\usepackage[preprint,nonatbib]{neurips_2021} %
\usepackage{times} %
\usepackage{helvet} %
\usepackage{courier} %
\usepackage[hyphens]{url} %
\usepackage{graphicx} %
\usepackage{amsmath}
\usepackage{booktabs}
\urlstyle{rm} %
\def\UrlFont{\rm}... | https://openreview.net/forum?id=H3NUh9Kft-c | https://arxiv.org/abs/2112.02656 | Please evaluate the proposed communication-efficient federated learning algorithms in our paper, specifically focusing on their effectiveness in reducing communication load and the impact on accuracy for NLP and vision tasks. |
Review of the paper
This paper considers an important and timely problem in federated learning on knowledge graphs (KGs), especially developing an attack model which incurs privacy leakage in an existing work, named FedE, and proposing a new privacy-preserving embedding aggregation framework to protect the privacy agai... |
\documentclass[11pt]{article}
\usepackage[]{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage[ruled,linesnumbered,vlined]{algorithm2e}
\SetAlFnt{\small}
\SetAlCapFnt{\small}
\SetAlCapNameFnt{\small}
\newcommand{\var}{\texttt}
\let\oldnl\nl%
\newco... | https://openreview.net/forum?id=raDf3qKzYb5 | https://arxiv.org/abs/2203.09553 | Please evaluate the paper based on its proposed attack model and defense mechanism in the context of federated learning on knowledge graphs, and provide feedback on the clarity of the system model, the global update equation, and any minor typos or errors you may have noticed. |
Review
This paper proposed a novel reconstruction attack method to infer the client's data utilizing the model updates in a knowledge graph setting. The proposed method achieves strong attacking accuracy on previous baselines. To overcome this, the authors further proposed a new method called FedR robust to the propose... |
\documentclass[11pt]{article}
\usepackage[]{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage[ruled,linesnumbered,vlined]{algorithm2e}
\SetAlFnt{\small}
\SetAlCapFnt{\small}
\SetAlCapNameFnt{\small}
\newcommand{\var}{\texttt}
\let\oldnl\nl%
\newco... | https://openreview.net/forum?id=raDf3qKzYb5 | https://arxiv.org/abs/2203.09553 | Please evaluate our paper, specifically addressing the following points: 1. How does the success rate of the proposed reconstruction attack method impact the privacy level of the randomized algorithm used in training the knowledge graph? 2. Can you assess the optimality of the attack evaluated in this work and discuss ... |
Good idea, but some inaccuracies
I think overall, this paper would be interesting for the workshop, and authors have proposed an interesting approach. There are issues in presentation and claims which can be improved and fixed. Hopefully, discussions at the workshop can help authors gather more feedback and continue th... |
\documentclass[11pt]{article}
\usepackage[]{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage[ruled,linesnumbered,vlined]{algorithm2e}
\SetAlFnt{\small}
\SetAlCapFnt{\small}
\SetAlCapNameFnt{\small}
\newcommand{\var}{\texttt}
\let\oldnl\nl%
\newco... | https://openreview.net/forum?id=raDf3qKzYb5 | https://arxiv.org/abs/2203.09553 | Please evaluate the paper based on the reviewer's feedback, focusing on the clarity of the proposed approach, the definition of privacy, and the coherence of the method presentation. |
Review on "Efficient Federated Learning on knowledge Graphs via Privacy-Preserving Relation Embedding Aggregation"
This work presents an extension of FedE, a recently proposed knowledge graph aggregation scheme in Federated Learning. Specifically, the authors tackle the privacy issue by aggregating relation embedding i... |
\documentclass[11pt]{article}
\usepackage[]{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage[ruled,linesnumbered,vlined]{algorithm2e}
\SetAlFnt{\small}
\SetAlCapFnt{\small}
\SetAlCapNameFnt{\small}
\newcommand{\var}{\texttt}
\let\oldnl\nl%
\newco... | https://openreview.net/forum?id=raDf3qKzYb5 | https://arxiv.org/abs/2203.09553 | Please evaluate the paper based on the experimental settings and the comparison to other related works, and provide a rating and confidence level for your evaluation. |
Overall nice work analyzing different ways to scale federated training of transformers
## Summary
This paper primarily studies the effect of partial variable training, quantization, and their combinations to enable training large language models in a federated setup. These techniques allow training large language model... | \documentclass[11pt]{article}
\usepackage{times}
\usepackage{latexsym}
\usepackage{amsmath}
\usepackage{fullpage}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\newcommand{\fix}{\marginpar{FIX}}
\newcommand{\new}{\marginpar{NEW}}
\newcommand{\fedavg}{\textsc{FedAvg}}
\usepackage{natbib}
... | https://openreview.net/forum?id=ShNG29KGF-c | https://arxiv.org/abs/2204.09715 | Evaluate the paper's analysis of different techniques for scaling up federated training of language models, including partial variable training and quantization, and provide suggestions for improvement. |
Official Review
**Summary Of The Paper:** This paper leverages several techniques for mitigating the communication and computation bottlenecks to train a Transformer in cross-device federated learning. They systematically evaluate partial model training, quantization, efficient transfer learning, and communication-effi... | \documentclass[11pt]{article}
\usepackage{times}
\usepackage{latexsym}
\usepackage{amsmath}
\usepackage{fullpage}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\newcommand{\fix}{\marginpar{FIX}}
\newcommand{\new}{\marginpar{NEW}}
\newcommand{\fedavg}{\textsc{FedAvg}}
\usepackage{natbib}
... | https://openreview.net/forum?id=ShNG29KGF-c | https://arxiv.org/abs/2204.09715 | Please evaluate the paper, focusing on its readability, structure, and the comprehensiveness of the experimental design. |
review
**Summary**
The paper studies cross-device federated learning for the problem of language modelling. The paper conducts a series of empirical studies to show that large and high-performing language models can be trained in the cross-device setting, with large Transformers partially fine-tuned through federated u... | \documentclass[11pt]{article}
\usepackage{times}
\usepackage{latexsym}
\usepackage{amsmath}
\usepackage{fullpage}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\newcommand{\fix}{\marginpar{FIX}}
\newcommand{\new}{\marginpar{NEW}}
\newcommand{\fedavg}{\textsc{FedAvg}}
\usepackage{natbib}
... | https://openreview.net/forum?id=ShNG29KGF-c | https://arxiv.org/abs/2204.09715 | Please evaluate the paper on cross-device federated learning for language modelling, specifically focusing on the empirical results and their practical value. |
Simple idea, strong empirical performance. Rigorous evaluation including interesting ablation analysis.
## Strengths
- Simple approach, trivial to implement
- Strong empirical performance
- Thorough empirical evaluation: multiple benchmarks (including skewed version of Sent140), ablation over token lengths and types of... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage[subtle]{savetrees}
\usepackage{multirow}
\usepackage{hyperref}
\usepackage{booktabs} %
\usepackage{tabularx}
\usepackage{grap... | https://openreview.net/forum?id=SawenqFzFb9 | https://arxiv.org/abs/2110.00135 | Please evaluate the paper based on its simple approach, strong empirical performance, thorough evaluation, and privacy-preserving nature. |
## Summary
This paper proposes learning a personalized sentiment analysis model from the text by appending or prepending a user-specific string (termed "UserIdentifier") to the input text. Then a single transformer model is finetuned on data from all the individuals. Incorporating user identifiers help learn a better... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage[subtle]{savetrees}
\usepackage{multirow}
\usepackage{hyperref}
\usepackage{booktabs} %
\usepackage{tabularx}
\usepackage{grap... | https://openreview.net/forum?id=SawenqFzFb9 | https://arxiv.org/abs/2110.00135 | Please evaluate the proposed method by comparing it with other baselines and discussing the strengths and weaknesses of the approach. |
review
**Summary**
The paper proposes a data augmentation method to handle personalized prediction for text classification problems.
**Overall comments**
The method is simple and seems to work well on the toy problems studied in the paper. The experiments are adequate for a workshop paper (but can be improved to pr... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage[subtle]{savetrees}
\usepackage{multirow}
\usepackage{hyperref}
\usepackage{booktabs} %
\usepackage{tabularx}
\usepackage{grap... | https://openreview.net/forum?id=SawenqFzFb9 | https://arxiv.org/abs/2110.00135 | Please provide your evaluation of the paper and its proposed method for personalized prediction in text classification. |
Effective method for conducting backdoor attack of federated NLP tasks
This paper introduces a practical approach for injecting backdoor attacks into a federated learned model. The attackers only manipulate the embedding layers of a model for injecting the backdoor. Compared to the previously proposed backdoor attack o... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amsmath}
\usepackage{enumitem}
\usepackage{adjustbox}
\usepackage{inconsolata}
\newcommand\nj[1]{\textcolor{black}{#1... | https://openreview.net/forum?id=S3ExnqKfF-9 | https://arxiv.org/abs/2204.14017 | Please evaluate the effectiveness of the proposed method for conducting backdoor attacks on federated NLP tasks in terms of its ease of injection and difficulty of detection compared to previous approaches. |
Interesting topic and the method can be further improved
This work proposes a practical backdoor attack against NLP models in the Federate Learning scenarios by inserting malicious tokens in the word embeddings. The author demonstrates its effectiveness through many practical scenarios, e.g., large trigger tokens, etc.... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amsmath}
\usepackage{enumitem}
\usepackage{adjustbox}
\usepackage{inconsolata}
\newcommand\nj[1]{\textcolor{black}{#1... | https://openreview.net/forum?id=S3ExnqKfF-9 | https://arxiv.org/abs/2204.14017 | Please evaluate the effectiveness and potential improvements of the proposed backdoor attack method in the context of Federated Learning for NLP models, considering factors such as the stealthiness of the backdoor triggers and the robustness against potential defense mechanisms. |
review
The paper presents a new attack extension of the embedding attack on NLP models to FL. Instead of training to optimize the whole model the attacker only focuses on a small single embedding of an unpopular token.
I really liked the idea and think that it has a good potential impact, however I have a couple of c... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amsmath}
\usepackage{enumitem}
\usepackage{adjustbox}
\usepackage{inconsolata}
\newcommand\nj[1]{\textcolor{black}{#1... | https://openreview.net/forum?id=S3ExnqKfF-9 | https://arxiv.org/abs/2204.14017 | Please evaluate the paper, focusing on the motivation, experiments, novelty, and potential impact of the proposed attack extension on Federated Learning for NLP models. |
Simple and elegant approach to post-hoc error correction
### Paper Summary:
This paper focuses on improving GPT-3's performance post-deployment, without any retraining, via a growing repository of interactive user feedback. Through correcting GPT-3's misunderstanding of question intent via a key-value store of user qu... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{array}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{booktabs}
\usepackage{microtype}
\newcommand{\eat}[1]{}
\newcommand{\red}[1]{\textcolor{red}{#1}}
\newcommand{\pc}[1]{\textc... | https://openreview.net/forum?id=Bx-fUfKedZ5 | https://arxiv.org/abs/2201.06009 | Please evaluate the effectiveness and contribution of the paper in improving GPT-3's performance post-deployment without retraining. |
Solid paper with small writing issues.
### Pros
* The paper presents a simple method that works.
* The proposed method does not require model re-training which would be expensive.
* The proposed method supports a natural user-machine interaction.
### Cons
* One downside of the paper is that it only studies the GPT-3... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{array}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{booktabs}
\usepackage{microtype}
\newcommand{\eat}[1]{}
\newcommand{\red}[1]{\textcolor{red}{#1}}
\newcommand{\pc}[1]{\textc... | https://openreview.net/forum?id=Bx-fUfKedZ5 | https://arxiv.org/abs/2201.06009 | Please evaluate the strengths and weaknesses of my paper, focusing on its methodology and contributions. |
A generally strong paper with some weaknesses to be addressed
The presented work proposes a straight-forward method for improving the performance of pre-trained LMs (PLMs) on a variety of tasks through corrective feedback. The feedback is first supplied by a (simulated) user and stored in a memory component. In the mai... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage{EMNLP2022}
\usepackage{array}
\usepackage{times}
\usepackage{latexsym}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{booktabs}
\usepackage{microtype}
\newcommand{\eat}[1]{}
\newcommand{\red}[1]{\textcolor{red}{#1}}
\newcommand{\pc}[1]{\textc... | https://openreview.net/forum?id=Bx-fUfKedZ5 | https://arxiv.org/abs/2201.06009 | Please evaluate the paper based on its proposed method for improving the performance of pre-trained language models through corrective feedback and the experimental evaluation provided. |
Review of the paper
# Summary
This work investigates the augmentation of pretrained language models (LMs) with knowledge graphs (KGs) for the cause-effect relation classification and commonsense causal reasoning tasks. They verbalize the ATOMC-2020 KG triples into natural language which they use to continually pretrain... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{acl}
\usepackage{times}
\usepackage{tabularx}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{makecell}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amssymb}%
\usepackage{pifont}%
\usepackage{booktabs}
\new... | https://openreview.net/forum?id=ShMlIzKgOW9 | https://arxiv.org/abs/2112.08615 | Please evaluate the paper based on its contributions in studying pretrained language models augmented with knowledge graphs for commonsense reasoning tasks, and its experimental results on cause-effect relation classification and commonsense causal reasoning benchmarks. |
Unique limited contribution of cause-effect models created by knowledge-augmented LM pretraining
This paper proposes a method for knowledge-augmented LM pretraining with cause-effect information. The method is targetted towards causal reasoning benchmarks (TCR, COPA, and BCOPA-CE). The method performs better than vanil... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{acl}
\usepackage{times}
\usepackage{tabularx}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{makecell}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amssymb}%
\usepackage{pifont}%
\usepackage{booktabs}
\new... | https://openreview.net/forum?id=ShMlIzKgOW9 | https://arxiv.org/abs/2112.08615 | Please evaluate the proposed method for knowledge-augmented LM pretraining with cause-effect information in terms of its novelty, clarity of citations, addressing the challenge of domain and downstream tasks, and the structure of the evaluation with benchmark-specific adaptations. |
Review
Summary
- The technique of continually pretraining language models on commonsense knowledge graph triples has been shown useful for some downstream tasks, but it may depend on the specific domain and tasks. This work investigates the effect of this technique on the task of cause-effect relation classification. T... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{acl}
\usepackage{times}
\usepackage{tabularx}
\usepackage{latexsym}
\usepackage{graphicx}
\usepackage{makecell}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{amssymb}%
\usepackage{pifont}%
\usepackage{booktabs}
\new... | https://openreview.net/forum?id=ShMlIzKgOW9 | https://arxiv.org/abs/2112.08615 | Please evaluate the effectiveness of our paper in investigating the impact of pretraining language models on commonsense knowledge graphs for cause-effect relation classification, including any weaknesses or areas for improvement that you may have identified. |
a good benchmark paper overall
The paper proposes CIKQA, a commonsense benchmark, which unifies several commonsense task into QA format and associates them with relevant knowledge. Experiments shows that models can better learn inference and generalize across tasks with proposed formulation and usage of knowledge.
Str... |
\documentclass[11pt,a4paper]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[TABBOTCAP]{subfigure}
\usepackage[shortlabels]{enumitem}
\usepackage{tikz-dependency}
\usepackage{algorithm}
\usepackage{algpseudocode}
\usepackage{multirow}
\usepackage{color}
\usepackage{helvet}
\usepackage{... | https://openreview.net/forum?id=S6Pl8ztg_b5 | https://arxiv.org/abs/2210.06246 | Please evaluate the paper based on its proposed benchmark, experiments, and analysis, and provide your overall assessment. |
Commonsense benchmark paper - simple and well-presented
Overall the paper seems comparatively complete and solid to me. The authors propose the benchmark CIKQA with a clear task formulation and detailed steps on how to extract supporting knowledge, as well as a strong baseline that takes advantage of its format. Experi... |
\documentclass[11pt,a4paper]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[TABBOTCAP]{subfigure}
\usepackage[shortlabels]{enumitem}
\usepackage{tikz-dependency}
\usepackage{algorithm}
\usepackage{algpseudocode}
\usepackage{multirow}
\usepackage{color}
\usepackage{helvet}
\usepackage{... | https://openreview.net/forum?id=S6Pl8ztg_b5 | https://arxiv.org/abs/2210.06246 | Please evaluate the paper based on its completeness, clarity of task formulation, steps for extracting supporting knowledge, baseline performance, experimental design, and novelty of the proposed benchmark CIKQA. |
review
The authors propose a reformulation of commonsense reasoning QA tasks
that attempts to separate knowledge (e.g., facts as specified by a KG)
from inference (i.e., reasoning over a given set of facts). Their
setup is to pair a small knowledge graph with each question that
contains the relevant knowledge to answer... |
\documentclass[11pt,a4paper]{article}
\usepackage{acl}
\usepackage{times}
\usepackage{latexsym}
\usepackage[TABBOTCAP]{subfigure}
\usepackage[shortlabels]{enumitem}
\usepackage{tikz-dependency}
\usepackage{algorithm}
\usepackage{algpseudocode}
\usepackage{multirow}
\usepackage{color}
\usepackage{helvet}
\usepackage{... | https://openreview.net/forum?id=S6Pl8ztg_b5 | https://arxiv.org/abs/2210.06246 | Please evaluate the proposed reformulation of commonsense reasoning QA tasks that separates knowledge from inference and incorporates a small knowledge graph with each question, based on the experimental results and the potential limitations mentioned in the review. |
Insightful Analysis on CCI
The authors present analysis on contextual commonsense inference (CCI) using GLUCOSE, a story dataset annotated with commonsense explanations. They argue that the conflation of CCI with language generation (used in original GLUCOSE task) hinders model performance and the evaluation protocol a... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{emnlp2021}
\usepackage{times}
\usepackage{latexsym}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{enumitem}
\newcommand{\textapprox}{\raisebox{0.5ex... | https://openreview.net/forum?id=Se-xHMYg_bc | https://arxiv.org/abs/2202.07880 | Please evaluate the paper based on its analysis of contextual commonsense inference (CCI) and the proposed CIS^2 model, focusing on the strengths and areas for improvement. |
Interesting hypothesis "Contextual Commonsense Inference should not be conflated with NLG as in GLUCOSE" but not much evidence.
What the paper is about: The paper argues that for contextual commonsense inference (commonsense understanding in some story), the GLUCOSE task conflates a different skill of natural language ... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{emnlp2021}
\usepackage{times}
\usepackage{latexsym}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{enumitem}
\newcommand{\textapprox}{\raisebox{0.5ex... | https://openreview.net/forum?id=Se-xHMYg_bc | https://arxiv.org/abs/2202.07880 | Please evaluate the effectiveness of our paper in proposing the CIS2 task as a better metric and task for contextual commonsense inference compared to the GLUCOSE task, providing evidence and potential alternative evaluation methods. |
The paper pinpointed a valid issue with the existing CCI task formulation. However, the findings are somewhat obvious, and the newly designed task might not generalize to settings where the CS inference is implicit and not part of the given story.
This paper critiques existing methods on Contextual Commonsense Inferenc... | \pdfoutput=1
\documentclass[11pt]{article}
\usepackage[]{emnlp2021}
\usepackage{times}
\usepackage{latexsym}
\usepackage{booktabs}
\usepackage{amsmath}
\usepackage{graphicx}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{microtype}
\usepackage{enumitem}
\newcommand{\textapprox}{\raisebox{0.5ex... | https://openreview.net/forum?id=Se-xHMYg_bc | https://arxiv.org/abs/2202.07880 | Evaluate the proposed task reformulation in the paper and discuss its potential limitations in handling cases where the inference output is not explicitly stated in the given story. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.