2020.05.04.txt

==========New Papers==========
1, TITLE:       Selecting Backtranslated Data from Multiple Sources for Improved Neural Machine Translation
http://arxiv.org/abs/2005.00308
AUTHORS:        Xabier Soto ; Dimitar Shterionov ; Alberto Poncelas ; Andy Way
HIGHLIGHT:      In this work we analyse the impact that data translated with rule-based, phrase-based statistical and neural MT systems has on new MT systems.

2, TITLE:       PCA-SRGAN: Incremental Orthogonal Projection Discrimination for Face Super-resolution
http://arxiv.org/abs/2005.00306
AUTHORS:        Hao Dou ; Chen Chen ; Xiyuan Hu ; Zhisen Hu ; Silong Peng
HIGHLIGHT:      To further improve the performance of GAN based models on super-resolving face images, we propose PCA-SRGAN which pays attention to the cumulative discrimination in the orthogonal projection space spanned by PCA projection matrix of face data.

3, TITLE:       Defocus Deblurring Using Dual-Pixel Data
http://arxiv.org/abs/2005.00305
AUTHORS:        Abdullah Abuolaim ; Michael S. Brown
HIGHLIGHT:      We propose an effective defocus deblurring method that exploits data available on dual-pixel (DP) sensors found on most modern cameras.

4, TITLE:       Can Multilingual Language Models Transfer to an Unseen Dialect? A Case Study on North African Arabizi
http://arxiv.org/abs/2005.00318
AUTHORS:        Benjamin Muller ; Benoit Sagot ; Djamé Seddah
HIGHLIGHT:      In this work, we study the ability of multilingual language models to process an unseen dialect.

5, TITLE:       Mind the Trade-off: Debiasing NLU Models without Degrading the In-distribution Performance
http://arxiv.org/abs/2005.00315
AUTHORS:        Prasetya Ajie Utama ; Nafise Sadat Moosavi ; Iryna Gurevych
COMMENTS:       to appear at ACL 2020
HIGHLIGHT:      In this paper, we address this trade-off by introducing a novel debiasing method, called confidence regularization, which discourage models from exploiting biases while enabling them to receive enough incentive to learn from all the training examples.

6, TITLE:       Self-supervised Knowledge Triplet Learning for Zero-shot Question Answering
http://arxiv.org/abs/2005.00316
AUTHORS:        Pratyay Banerjee ; Chitta Baral
COMMENTS:       5 pages, 1 figure, 2 tables, WIP
HIGHLIGHT:      In this work, we propose Knowledge Triplet Learning, a self-supervised task over knowledge graphs.

7, TITLE:       Language (Re)modelling: Towards Embodied Language Understanding
http://arxiv.org/abs/2005.00311
AUTHORS:        Ronen Tamari ; Chen Shani ; Tom Hope ; Miriam R. L. Petruck ; Omri Abend ; Dafna Shahaf
COMMENTS:       Accepted to ACL2020 Theme Track
HIGHLIGHT:      This work proposes an approach to representation and learning based on the tenets of embodied cognitive linguistics (ECL).

8, TITLE:       CDL: Curriculum Dual Learning for Emotion-Controllable Response Generation
http://arxiv.org/abs/2005.00329
AUTHORS:        Lei Shen ; Yang Feng
COMMENTS:       To appear at ACL 2020 (long paper)
HIGHLIGHT:      To alleviate these problems, we propose a novel framework named Curriculum Dual Learning (CDL) which extends the emotion-controllable response generation to a dual task to generate emotional responses and emotional queries alternatively.

9, TITLE:       ACCL: Adversarial constrained-CNN loss for weakly supervised medical image segmentation
http://arxiv.org/abs/2005.00328
AUTHORS:        Pengyi Zhang ; Yunxin Zhong ; Xiaoqiong Li
HIGHLIGHT:      We propose adversarial constrained-CNN loss, a new paradigm of constrained-CNN loss methods, for weakly supervised medical image segmentation.

10, TITLE:      Diverse Visuo-Lingustic Question Answering (DVLQA) Challenge
http://arxiv.org/abs/2005.00330
AUTHORS:        Shailaja Sampat ; Yezhou Yang ; Chitta Baral
COMMENTS:       12 pages, 6 figures
HIGHLIGHT:      Diverse Visuo-Lingustic Question Answering (DVLQA) Challenge

11, TITLE:      XCOPA: A Multilingual Dataset for Causal Commonsense Reasoning
http://arxiv.org/abs/2005.00333
AUTHORS:        Edoardo Maria Ponti ; Goran Glavaš ; Olga Majewska ; Qianchu Liu ; Ivan Vulić ; Anna Korhonen
HIGHLIGHT:      Motivated by both demands, we introduce Cross-lingual Choice of Plausible Alternatives (XCOPA), a typologically diverse multilingual dataset for causal commonsense reasoning in 11 languages.

12, TITLE:      Adversarial Synthesis of Human Pose from Text
http://arxiv.org/abs/2005.00340
AUTHORS:        Yifei Zhang ; Rania Briq ; Julian Tanke ; Juergen Gall
HIGHLIGHT:      In order to solve this task, we propose a model that is based on a conditional generative adversarial network.

13, TITLE:      Linguistic Typology Features from Text: Inferring the Sparse Features of World Atlas of Language Structures
http://arxiv.org/abs/2005.00100
AUTHORS:        Alexander Gutkin ; Tatiana Merkulova ; Martin Jansche
COMMENTS:       Originally prepared as a conference submission to EMNLP 2018
HIGHLIGHT:      In this paper we investigate whether the various linguistic features from World Atlas of Language Structures (WALS) can be reliably inferred from multi-lingual text.

14, TITLE:      Multilingual Unsupervised Sentence Simplification
http://arxiv.org/abs/2005.00352
AUTHORS:        Louis Martin ; Angela Fan ; Éric de la Clergerie ; Antoine Bordes ; Benoît Sagot
HIGHLIGHT:      In this work, we propose using unsupervised mining techniques to automatically create training corpora for simplification in multiple languages from raw Common Crawl web data.

15, TITLE:      On the Spontaneous Emergence of Discrete and Compositional Signals
http://arxiv.org/abs/2005.00110
AUTHORS:        Nur Geffen Lan ; Emmanuel Chemla ; Shane Steinert-Threlkeld
COMMENTS:       ACL 2020
HIGHLIGHT:      We propose a general framework to study language emergence through signaling games with neural agents.

16, TITLE:      Sequence Information Channel Concatenation for Improving Camera Trap Image Burst Classification
http://arxiv.org/abs/2005.00116
AUTHORS:        Bhuvan Malladihalli Shashidhara ; Darshan Mehta ; Yash Kale ; Dan Morris ; Megan Hazen
COMMENTS:       9 pages, 4 figures, 2 tables. Git repository can be found at: https://github.com/bhuvi3/camera_trap_animal_classification
HIGHLIGHT:      In this work, we explore a variety of approaches, and measure the impact of using short image sequences (burst of 3 images) on improving the camera trap image classification.

17, TITLE:      Learning to Rank Intents in Voice Assistants
http://arxiv.org/abs/2005.00119
AUTHORS:        Raviteja Anantha ; Srinivas Chappidi ; William Dawoodi
COMMENTS:       11 pages, 7 figures, 2 tables, accepted at IWSDS 2020 conference
HIGHLIGHT:      In this work, we propose a novel Energy-based model for the intent ranking task, where we learn an affinity metric and model the trade-off between extracted meaning from speech utterances and relevance/executability aspects of the intent.

18, TITLE:      Beneath the Tip of the Iceberg: Current Challenges and New Directions in Sentiment Analysis Research
http://arxiv.org/abs/2005.00357
AUTHORS:        Soujanya Poria ; Devamanyu Hazarika ; Navonil Majumder ; Rada Mihalcea
HIGHLIGHT:      In this article, we discuss this perception by pointing out the shortcomings and under-explored, yet key aspects of this field that are necessary to attain true sentiment understanding.

19, TITLE:      Learning to Faithfully Rationalize by Construction
http://arxiv.org/abs/2005.00115
AUTHORS:        Sarthak Jain ; Sarah Wiegreffe ; Yuval Pinter ; Byron C. Wallace
COMMENTS:       ACL2020 Camera Ready Submission
HIGHLIGHT:      Lei et al. (2016) proposed a model to produce faithful rationales for neural text classification by defining independent snippet extraction and prediction modules.

20, TITLE:      A Naturalness Evaluation Database for Video Prediction Models
http://arxiv.org/abs/2005.00356
AUTHORS:        Nagabhushan Somraj ; Manoj Surya Kashi ; S. P. Arun ; Rajiv Soundararajan
COMMENTS:       Project website: https://sites.google.com/site/nagabhushansn95/publications/vine
HIGHLIGHT:      In this context, we introduce the problem of naturalness evaluation, which refers to how natural or realistic a predicted video looks.

21, TITLE:      Revisiting Memory-Efficient Incremental Coreference Resolution
http://arxiv.org/abs/2005.00128
AUTHORS:        Patrick Xia ; João Sedoc ; Benjamin Van Durme
HIGHLIGHT:      We explore the task of coreference resolution under fixed memory by extending an incremental clustering algorithm to utilize contextualized encoders and neural components.

22, TITLE:      Conceptual Design of Human-Drone Communication in Collaborative Environments
http://arxiv.org/abs/2005.00127
AUTHORS:        Hans Dermot Doran ; Monika Reif ; Marco Oehler ; Curdin Stoehr ; Pierluigi Capone
COMMENTS:       4 pages, 4 figures
HIGHLIGHT:      We present basic visual indicators enhanced with flight patterns for drone-human interaction and human signaling based on aircraft marshaling for humane-drone interaction.

23, TITLE:      Structure-Tags Improve Text Classification for Scholarly Document Quality Prediction
http://arxiv.org/abs/2005.00129
AUTHORS:        Gideon Maillette de Buy Wenniger ; Thomas van Dongen ; Eleri Aedmaa ; Herbert Teun Kruitbosch ; Edwin A. Valentijn ; Lambert Schomaker
HIGHLIGHT:      To tackle these problems, we propose the use of HANs combined with structure-tags which mark the role of sentences in the document.

24, TITLE:      Unsupervised Learning of KB Queries in Task Oriented Dialogs
http://arxiv.org/abs/2005.00123
AUTHORS:        Dinesh Raghu ; Nikhil Gupta ;  Mausam
HIGHLIGHT:      In this paper, we propose a novel problem of learning end-to-end TOD systems using dialogs that do not contain KB query annotations.

25, TITLE:      Hide-and-Seek: A Template for Explainable AI
http://arxiv.org/abs/2005.00130
AUTHORS:        Thanos Tagaris ; Andreas Stafylopatis
COMMENTS:       24 pages, 14 figures. Submitted on a special issue for Explainable AI, on Elsevier's "Artificial Intelligence"
HIGHLIGHT:      This study proposes a novel framework called Hide-and-Seek (HnS) for training Interpretable Neural Networks and establishes a theoretical foundation for exploring and comparing similar ideas.

26, TITLE:      A Comprehensive Study on Visual Explanations for Spatio-temporal Networks
http://arxiv.org/abs/2005.00375
AUTHORS:        Zhenqiang Li ; Weimin Wang ; Zuoyue Li ; Yifei Huang ; Yoichi Sato
HIGHLIGHT:      In this paper, we provide a comprehensive study of the existing video attribution methods of two categories, gradient-based and perturbation-based, for visual explanation of neural networks that take videos as the input (spatio-temporal networks).

27, TITLE:      Contextual Text Style Transfer
http://arxiv.org/abs/2005.00136
AUTHORS:        Yu Cheng ; Zhe Gan ; Yizhe Zhang ; Oussama Elachqar ; Dianqi Li ; Jingjing Liu
HIGHLIGHT:      To realize high-quality style transfer with natural context preservation, we propose a Context-Aware Style Transfer (CAST) model, which uses two separate encoders for each input sentence and its surrounding context.

28, TITLE:      MOPS-Net: A Matrix Optimization-driven Network forTask-Oriented 3D Point Cloud Downsampling
http://arxiv.org/abs/2005.00383
AUTHORS:        Yue Qian ; Junhui Hou ; Yiming Zeng ; Qijian Zhang ; Sam Kwong ; Ying He
COMMENTS:       12 pages, 11 figures, 7 tables
HIGHLIGHT:      We propose MOPS-Net, a novel end-to-end deepneural network which is designed from the perspective of matrixoptimization, making it fundamentally different from the existing deep learning-based methods.

29, TITLE:      Will-They-Won't-They: A Very Large Dataset for Stance Detection on Twitter
http://arxiv.org/abs/2005.00388
AUTHORS:        Costanza Conforti ; Jakob Berndt ; Mohammad Taher Pilehvar ; Chryssi Giannitsarou ; Flavio Toxvaerd ; Nigel Collier
COMMENTS:       10 pages, accepted at ACL2020
HIGHLIGHT:      We present a new challenging stance detection dataset, called Will-They-Won't-They (WT-WT), which contains 51,284 tweets in English, making it by far the largest available dataset of the type.

30, TITLE:      Interpretable Entity Representations through Large-Scale Typing
http://arxiv.org/abs/2005.00147
AUTHORS:        Yasumasa Onoe ; Greg Durrett
HIGHLIGHT:      In this paper, we present an approach to creating interpretable entity representations that are human readable and achieve high performance on entity-related tasks out of the box.

31, TITLE:      Enriching Documents with Compact, Representative, Relevant Knowledge Graphs
http://arxiv.org/abs/2005.00153
AUTHORS:        Shuxin Li ; Zixian Huang ; Gong Cheng ; Evgeny Kharlamov ; Kalpa Gunaratna
COMMENTS:       7 pages, accepted to IJCAI-PRICAI 2020
HIGHLIGHT:      Existing methods identify mentions of entities in a background KG and enrich documents with entity types and direct relations.

32, TITLE:      Neural Entity Summarization with Joint Encoding and Weak Supervision
http://arxiv.org/abs/2005.00152
AUTHORS:        Junyou Li ; Gong Cheng ; Qingxia Liu ; Wen Zhang ; Evgeny Kharlamov ; Kalpa Gunaratna ; Huajun Chen
COMMENTS:       7 pages, accepted to IJCAI-PRICAI 2020
HIGHLIGHT:      In this paper, we present a supervised approach NEST that is based on our novel neural model to jointly encode graph structure and text in KGs and generate high-quality diversified summaries.

33, TITLE:      Identifying Necessary Elements for BERT's Multilinguality
http://arxiv.org/abs/2005.00396
AUTHORS:        Philipp Dufter ; Hinrich Schütze
HIGHLIGHT:      We aim to identify architectural properties of BERT as well as linguistic properties of languages that are necessary for BERT to become multilingual.

34, TITLE:      Why and when should you pool? Analyzing Pooling in Recurrent Architectures
http://arxiv.org/abs/2005.00159
AUTHORS:        Pratyush Maini ; Keshav Kolluru ; Danish Pruthi ;  Mausam
COMMENTS:       Preprint
HIGHLIGHT:      In this work, we examine three commonly used pooling techniques (mean-pooling, max-pooling, and attention), and propose max-attention, a novel variant that effectively captures interactions among predictive tokens in a sentence.

35, TITLE:      Attend to Medical Ontologies: Content Selection for Clinical Abstractive Summarization
http://arxiv.org/abs/2005.00163
AUTHORS:        Sajad Sotudeh ; Nazli Goharian ; Ross W. Filice
COMMENTS:       Accepted to ACL 2020
HIGHLIGHT:      In this paper, we approach the content selection problem for clinical abstractive summarization by augmenting salient ontological terms into the summarizer.

36, TITLE:      Recurrent Neural Network Language Models Always Learn English-Like Relative Clause Attachment
http://arxiv.org/abs/2005.00165
AUTHORS:        Forrest Davis ; Marten van Schijndel
COMMENTS:       Proceedings of 58th Annual Meeting of the Association for Computational Linguistics
HIGHLIGHT:      A standard approach to evaluating language models analyzes how models assign probabilities to valid versus invalid syntactic constructions (i.e. is a grammatical sentence more probable than an ungrammatical sentence).

37, TITLE:      Recurrent Interaction Network for Jointly Extracting Entities and Classifying Relations
http://arxiv.org/abs/2005.00162
AUTHORS:        Kai Sun ; Richong Zhang ; Samuel Mensah ; Yongyi Mao ; Xudong Liu
HIGHLIGHT:      In this study, we argue that an explicit interaction between the NER model and the RE model will better guide the training of both models.

38, TITLE:      Selecting Informative Contexts Improves Language Model Finetuning
http://arxiv.org/abs/2005.00175
AUTHORS:        Richard Antonello ; Javier Turek ; Alexander Huth
HIGHLIGHT:      We present a general finetuning meta-method that we call information gain filtration for improving the overall training efficiency and final performance of language model finetuning.

39, TITLE:      Universal Adversarial Attacks with Natural Triggers for Text Classification
http://arxiv.org/abs/2005.00174
AUTHORS:        Liwei Song ; Xinwei Yu ; Hsuan-Tung Peng ; Karthik Narasimhan
COMMENTS:       code is available at https://github.com/Hsuan-Tung/universal_attack_natural_trigger
HIGHLIGHT:      In this paper, we develop adversarial attacks that appear closer to natural English phrases and yet confuse classification systems when added to benign inputs.

40, TITLE:      Cross-lingual Entity Alignment for Knowledge Graphs with Incidental Supervision from Free Text
http://arxiv.org/abs/2005.00171
AUTHORS:        Muhao Chen ; Weijia Shi ; Ben Zhou ; Dan Roth
HIGHLIGHT:      Therefore, we propose a new model, JEANS , which jointly represents multilingual KGs and text corpora in a shared embedding scheme, and seeks to improve entity alignment with incidental supervision signals from text.

41, TITLE:      Information Seeking in the Spirit of Learning: a Dataset for Conversational Curiosity
http://arxiv.org/abs/2005.00172
AUTHORS:        Pedro Rodriguez ; Paul Crook ; Seungwhan Moon ; Zhiguang Wang
HIGHLIGHT:      We incorporate this knowledge into a state-of-the-art multi-task model that reproduces human assistant policies, improving over content selection baselines by 13 points.

42, TITLE:      Cross-Linguistic Syntactic Evaluation of Word Prediction Models
http://arxiv.org/abs/2005.00187
AUTHORS:        Aaron Mueller ; Garrett Nicolai ; Panayiota Petrou-Zeniou ; Natalia Talmina ; Tal Linzen
COMMENTS:       Accepted for presentation at ACL 2020
HIGHLIGHT:      To investigate how these models' ability to learn syntax varies by language, we introduce CLAMS (Cross-Linguistic Assessment of Models on Syntax), a syntactic evaluation suite for monolingual and multilingual models.

43, TITLE:      Sparse, Dense, and Attentional Representations for Text Retrieval
http://arxiv.org/abs/2005.00181
AUTHORS:        Yi Luan ; Jacob Eisenstein ; Kristina Toutanova ; Michael Collins
HIGHLIGHT:      Building on these insights, we propose a simple neural model that combines the efficiency of dual encoders with some of the expressiveness of attentional architectures, and explore a sparse-dense hybrid to capitalize on the precision of sparse retrieval.

44, TITLE:      Efficient lambda encodings for Mendler-style coinductive types in Cedille
http://arxiv.org/abs/2005.00199
AUTHORS:        Christopher Jenkins ; Aaron Stump ; Larry Diehl
COMMENTS:       In Proceedings MSFP 2020, arXiv:2004.14735
HIGHLIGHT:      All work is mechanically verified by the Cedille tool, an implementation of CDLE.

45, TITLE:      Multi-dimensional Arrays with Levels
http://arxiv.org/abs/2005.00198
AUTHORS:        Artjoms {Š}inkarovs
COMMENTS:       In Proceedings MSFP 2020, arXiv:2004.14735
HIGHLIGHT:      In this paper we present an Agda formalisation of arrays with levels.

46, TITLE:      KPQA: A Metric for Generative Question Answering Using Word Weights
http://arxiv.org/abs/2005.00192
AUTHORS:        Hwanhee Lee ; Seunghyun Yoon ; Franck Dernoncourt ; Doo Soon Kim ; Trung Bui ; Joongbo Shin ; Kyomin Jung
HIGHLIGHT:      To alleviate this problem, we propose a new metric for evaluating the correctness of genQA.

47, TITLE:      Evaluating Neural Machine Comprehension Model Robustness to Noisy Inputs and Adversarial Attacks
http://arxiv.org/abs/2005.00190
AUTHORS:        Winston Wu ; Dustin Arendt ; Svitlana Volkova
HIGHLIGHT:      We evaluate machine comprehension models' robustness to noise and adversarial attacks by performing novel perturbations at the character, word, and sentence level.

48, TITLE:      Partially-Typed NER Datasets Integration: Connecting Practice to Theory
http://arxiv.org/abs/2005.00502
AUTHORS:        Shi Zhi ; Liyuan Liu ; Yu Zhang ; Shiyin Wang ; Qi Li ; Chao Zhang ; Jiawei Han
COMMENTS:       Work in progress
HIGHLIGHT:      Here, we conduct a systematic analysis and comparison between partially-typed NER datasets and fully-typed ones, in both theoretical and empirical manner.

49, TITLE:      HipoRank: Incorporating Hierarchical and Positional Information into Graph-based Unsupervised Long Document Extractive Summarization
http://arxiv.org/abs/2005.00513
AUTHORS:        Yue Dong ; Andrei Romascanu ; Jackie C. K. Cheung
COMMENTS:       9 pages, 3 figures
HIGHLIGHT:      We propose a novel graph-based ranking model for unsupervised extractive summarization of long documents.

50, TITLE:      SciREX: A Challenge Dataset for Document-Level Information Extraction
http://arxiv.org/abs/2005.00512
AUTHORS:        Sarthak Jain ; Madeleine van Zuylen ; Hannaneh Hajishirzi ; Iz Beltagy
COMMENTS:       ACL2020 Camera Ready Submission, Work done by first authors while interning at AI2
HIGHLIGHT:      In this paper, we introduce SciREX, a document level IE dataset that encompasses multiple IE tasks, including salient entity identification and document level $N$-ary relation identification from scientific articles. We annotate our dataset by integrating automatic and human annotations, leveraging existing scientific knowledge resources.

51, TITLE:      Is Long Horizon Reinforcement Learning More Difficult Than Short Horizon Reinforcement Learning?
http://arxiv.org/abs/2005.00527
AUTHORS:        Ruosong Wang ; Simon S. Du ; Lin F. Yang ; Sham M. Kakade
HIGHLIGHT:      Our analysis introduces two ideas: (i) the construction of an $\varepsilon$-net for optimal policies whose log-covering number scales only logarithmically with the planning horizon, and (ii) the Online Trajectory Synthesis algorithm, which adaptively evaluates all policies in a given policy class using sample complexity that scales with the log-covering number of the given policy class.

52, TITLE:      Why Overfitting Isn't Always Bad: Retrofitting Cross-Lingual Word Embeddings to Dictionaries
http://arxiv.org/abs/2005.00524
AUTHORS:        Mozhi Zhang ; Yoshinari Fujinuma ; Michael J. Paul ; Jordan Boyd-Graber
COMMENTS:       ACL 2020
HIGHLIGHT:      We address this limitation by retrofitting CLWE to the training dictionary, which pulls training translation pairs closer in the embedding space and overfits the training dictionary.

53, TITLE:      Aggregation and Finetuning for Clothes Landmark Detection
http://arxiv.org/abs/2005.00419
AUTHORS:        Tzu-Heng Lin
COMMENTS:       Technical report, 4 pages
HIGHLIGHT:      In this paper, a new training scheme for clothes landmark detection: $\textit{Aggregation and Finetuning}$, is proposed.

54, TITLE:      Investigating Class-level Difficulty Factors in Multi-label Classification Problems
http://arxiv.org/abs/2005.00430
AUTHORS:        Mark Marsden ; Kevin McGuinness ; Joseph Antony ; Haolin Wei ; Milan Redzic ; Jian Tang ; Zhilan Hu ; Alan Smeaton ; Noel E O'Connor
COMMENTS:       Published in ICME 2020
HIGHLIGHT:      This work investigates the use of class-level difficulty factors in multi-label classification problems for the first time.

55, TITLE:      Bipartite Flat-Graph Network for Nested Named Entity Recognition
http://arxiv.org/abs/2005.00436
AUTHORS:        Ying Luo ; Hai Zhao
COMMENTS:       Accepted by ACL2020
HIGHLIGHT:      In this paper, we propose a novel bipartite flat-graph network (BiFlaG) for nested named entity recognition (NER), which contains two subgraph modules: a flat NER module for outermost entities and a graph module for all the entities located in inner layers.

56, TITLE:      Topological Sort for Sentence Ordering
http://arxiv.org/abs/2005.00432
AUTHORS:        Shrimai Prabhumoye ; Ruslan Salakhutdinov ; Alan W Black
COMMENTS:       Will be published at the Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL) 2020
HIGHLIGHT:      In this paper, we propose a new framing of this task as a constraint solving problem and introduce a new technique to solve it.

57, TITLE:      Multi-head Monotonic Chunkwise Attention For Online Speech Recognition
http://arxiv.org/abs/2005.00205
AUTHORS:        Baiji Liu ; Songjun Cao ; Sining Sun ; Weibin Zhang ; Long Ma
HIGHLIGHT:      To deal with this problem, we propose multi-head monotonic chunk-wise attention (MTH-MoChA), an improved version of MoChA.

58, TITLE:      Defense of Word-level Adversarial Attacks via Random Substitution Encoding
http://arxiv.org/abs/2005.00446
AUTHORS:        Zhaoyang Wang ; Hongtao Wang
COMMENTS:       12 pages, 2 figures, 4 tables
HIGHLIGHT:      In this paper, we shed light on this problem and propose a novel defense framework called Random Substitution Encoding (RSE), which introduces a random substitution encoder into the training process of original neural networks.

59, TITLE:      TransOMCS: From Linguistic Graphs to Commonsense Knowledge
http://arxiv.org/abs/2005.00206
AUTHORS:        Hongming Zhang ; Daniel Khashabi ; Yangqiu Song ; Dan Roth
COMMENTS:       Accepted by IJCAI 2020
HIGHLIGHT:      In this paper, we explore a practical way of mining commonsense knowledge from linguistic graphs, with the goal of transferring cheap knowledge obtained with linguistic patterns into expensive commonsense knowledge.

60, TITLE:      HERO: Hierarchical Encoder for Video+Language Omni-representation Pre-training
http://arxiv.org/abs/2005.00200
AUTHORS:        Linjie Li ; Yen-Chun Chen ; Yu Cheng ; Zhe Gan ; Licheng Yu ; Jingjing Liu
HIGHLIGHT:      We present HERO, a Hierarchical EncodeR for Omni-representation learning, for large-scale video+language pre-training.

61, TITLE:      Computing the Testing Error without a Testing Set
http://arxiv.org/abs/2005.00450
AUTHORS:        Ciprian Corneanu ; Meysam Madadi ; Sergio Escalera ; Aleix Martinez
HIGHLIGHT:      Here, we derive an algorithm to estimate the performance gap between training and testing that does not require any testing dataset.

62, TITLE:      Style Variation as a Vantage Point for Code-Switching
http://arxiv.org/abs/2005.00458
AUTHORS:        Khyathi Raghavi Chandu ; Alan W Black
HIGHLIGHT:      We present a novel vantage point of CS to be style variations between both the participating languages.

63, TITLE:      USR: An Unsupervised and Reference Free Evaluation Metric for Dialog Generation
http://arxiv.org/abs/2005.00456
AUTHORS:        Shikib Mehri ; Maxine Eskenazi
COMMENTS:       Accepted to ACL 2020 as long paper
HIGHLIGHT:      To this end, this paper presents USR, an UnSupervised and Reference-free evaluation metric for dialog.

64, TITLE:      The AVA-Kinetics Localized Human Actions Video Dataset
http://arxiv.org/abs/2005.00214
AUTHORS:        Ang Li ; Meghana Thotakuri ; David A. Ross ; João Carreira ; Alexander Vostrikov ; Andrew Zisserman
COMMENTS:       8 pages, 8 figures
HIGHLIGHT:      This paper describes the AVA-Kinetics localized human actions video dataset.

65, TITLE:      MedType: Improving Medical Entity Linking with Semantic Type Prediction
http://arxiv.org/abs/2005.00460
AUTHORS:        Shikhar Vashishth ; Rishabh Joshi ; Ritam Dutt ; Denis Newman-Griffis ; Carolyn Rose
COMMENTS:       14 pages
HIGHLIGHT:      In this paper, we probe the impact of incorporating an entity disambiguation step in existing entity linkers.

66, TITLE:      HLVU : A New Challenge to Test Deep Understanding of Movies the Way Humans do
http://arxiv.org/abs/2005.00463
AUTHORS:        Keith Curtis ; George Awad ; Shahzad Rajput ; Ian Soboroff
HIGHLIGHT:      In this paper we propose a new evaluation challenge and direction in the area of High-level Video Understanding.

67, TITLE:      Automatic Discourse Segmentation: Review and Perspectives
http://arxiv.org/abs/2005.00468
AUTHORS:        Iria da Cunha ; Juan-Manuel Torres-Moreno
COMMENTS:       5 pages, 1 figure
HIGHLIGHT:      The study reported in this article addresses a review of two on-line available discourse segmenters (for English and Portuguese).

68, TITLE:      Deepfake Forensics Using Recurrent Neural Networks
http://arxiv.org/abs/2005.00229
AUTHORS:        Rahul U ; Ragul M ; Raja Vignesh K ; Tejeswinee K
HIGHLIGHT:      This paper proposes a transient mindful pipeline to automat-ically recognize deepfake recordings.

69, TITLE:      Deeply Cascaded U-Net for Multi-Task Image Processing
http://arxiv.org/abs/2005.00225
AUTHORS:        Ilja Gubins ; Remco C. Veltkamp
COMMENTS:       6 pages, 6 figures
HIGHLIGHT:      In this paper, we propose a novel multi-task neural network architecture designed for combining sequential image processing tasks.

70, TITLE:      Biomedical Entity Representations with Synonym Marginalization
http://arxiv.org/abs/2005.00239
AUTHORS:        Mujeen Sung ; Hwisang Jeon ; Jinhyuk Lee ; Jaewoo Kang
COMMENTS:       ACL 2020
HIGHLIGHT:      In this paper, we focus on learning representations of biomedical entities solely based on the synonyms of entities.

71, TITLE:      TORQUE: A Reading Comprehension Dataset of Temporal Ordering Questions
http://arxiv.org/abs/2005.00242
AUTHORS:        Qiang Ning ; Hao Wu ; Rujun Han ; Nanyun Peng ; Matt Gardner ; Dan Roth
COMMENTS:       15 pages (incl. 4 pages in the appendix)
HIGHLIGHT:      We introduce TORQUE, a new English reading comprehension benchmark built on 3.2k news snippets with 21k human-generated questions querying temporal relationships.

72, TITLE:      ASSET: A Dataset for Tuning and Evaluation of Sentence Simplification Models with Multiple Rewriting Transformations
http://arxiv.org/abs/2005.00481
AUTHORS:        Fernando Alva-Manchego ; Louis Martin ; Antoine Bordes ; Carolina Scarton ; Benoît Sagot ; Lucia Specia
COMMENTS:       Accepted to ACL 2020 (camera-ready version)
HIGHLIGHT:      To alleviate this limitation, this paper introduces ASSET, a new dataset for assessing sentence simplification in English.

73, TITLE:      Knowledge Base Inference for Regular Expression Queries
http://arxiv.org/abs/2005.00480
AUTHORS:        Vaibhav Adlakha ; Parth Shah ; Srikanta Bedathur ;  Mausam
HIGHLIGHT:      In response, we present Regex Query Answering, the novel task of answering regex queries on incomplete KBs.

74, TITLE:      AdapterFusion: Non-Destructive Task Composition for Transfer Learning
http://arxiv.org/abs/2005.00247
AUTHORS:        Jonas Pfeiffer ; Aishwarya Kamath ; Andreas Rücklé ; Kyunghyun Cho ; Iryna Gurevych
HIGHLIGHT:      We propose a new architecture as well as a two-stage learning algorithm that allows us to effectively share knowledge from multiple tasks while avoiding these crucial problems.

75, TITLE:      Cross-modal Language Generation using Pivot Stabilization for Web-scale Language Coverage
http://arxiv.org/abs/2005.00246
AUTHORS:        Ashish V. Thapliyal ; Radu Soricut
COMMENTS:       ACL 2020
HIGHLIGHT:      We describe an approach called Pivot-Language Generation Stabilization (PLuGS), which leverages directly at training time both existing English annotations (gold data) as well as their machine-translated versions (silver data); at run-time, it generates first an English caption and then a corresponding target-language caption.

76, TITLE:      Structured Tuning for Semantic Role Labeling
http://arxiv.org/abs/2005.00496
AUTHORS:        Tao Li ; Parth Anand Jawale ; Martha Palmer ; Vivek Srikumar
COMMENTS:       Accepted at ACL 2020
HIGHLIGHT:      In this paper, we present a structured tuning framework to improve models using softened constraints only at training time.

77, TITLE:      Recognizing American Sign Language Nonmanual Signal Grammar Errors in Continuous Videos
http://arxiv.org/abs/2005.00253
AUTHORS:        Elahe Vahdani ; Longlong Jing ; Yingli Tian ; Matt Huenerfauth
HIGHLIGHT:      As part of the development of an educational tool that can help students achieve fluency in American Sign Language (ASL) through independent and interactive practice with immediate feedback, this paper introduces a near real-time system to recognize grammatical errors in continuous signing videos without necessarily identifying the entire sequence of signs. We have collected a dataset of continuous sign language, ASL-HW-RGBD, covering different aspects of ASL grammars for training and testing.

78, TITLE:      Low Resource Multi-Task Sequence Tagging -- Revisiting Dynamic Conditional Random Fields
http://arxiv.org/abs/2005.00250
AUTHORS:        Jonas Pfeiffer ; Edwin Simpson ; Iryna Gurevych
HIGHLIGHT:      We compare different models for low resource multi-task sequence tagging that leverage dependencies between label sequences for different tasks.

79, TITLE:      Representations of Syntax [MASK] Useful: Effects of Constituency and Dependency Structure in Recursive LSTMs
http://arxiv.org/abs/2005.00019
AUTHORS:        Michael A. Lepori ; Tal Linzen ; R. Thomas McCoy
COMMENTS:       To appear in Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics (ACL-2020)
HIGHLIGHT:      We evaluate which of these two representational schemes more effectively introduces biases for syntactic structure that increase performance on the subject-verb agreement prediction task.

80, TITLE:      An Efficient Integration of Disentangled Attended Expression and Identity FeaturesFor Facial Expression Transfer andSynthesis
http://arxiv.org/abs/2005.00499
AUTHORS:        Kamran Ali ; Charles E. Hughes
COMMENTS:       10 Pages, excluding references
HIGHLIGHT:      In this paper, we present an Attention-based Identity Preserving Generative Adversarial Network (AIP-GAN) to overcome the identity leakage problem from a source image to a generated face image, an issue that is encountered in a cross-subject facial expression transfer and synthesis process.

81, TITLE:      Towards Controllable Biases in Language Generation
http://arxiv.org/abs/2005.00268
AUTHORS:        Emily Sheng ; Kai-Wei Chang ; Premkumar Natarajan ; Nanyun Peng
COMMENTS:       9 pages
HIGHLIGHT:      We present a general approach towards controllable societal biases in natural language generation (NLG).

82, TITLE:      Unsupervised Lesion Detection via Image Restoration with a Normative Prior
http://arxiv.org/abs/2005.00031
AUTHORS:        Xiaoran Chen ; Suhang You ; Kerem Can Tezcan ; Ender Konukoglu
COMMENTS:       Extended version of 'Unsupervised Lesion Detection via Image Restoration with a Normative Prior' (MIDL2019)
HIGHLIGHT:      In this work, we approach unsupervised lesion detection as an image restoration problem and propose a probabilistic model that uses a network-based prior as the normative distribution and detect lesions pixel-wise using MAP estimation.

83, TITLE:      Fighting the COVID-19 Infodemic: Modeling the Perspective of Journalists, Fact-Checkers, Social Media Platforms, Policy Makers, and the Society
http://arxiv.org/abs/2005.00033
AUTHORS:        Firoj Alam ; Shaden Shaar ; Alex Nikolov ; Hamdy Mubarak ; Giovanni Da San Martino ; Ahmed Abdelali ; Fahim Dalvi ; Nadir Durrani ; Hassan Sajjad ; Kareem Darwish ; Preslav Nakov
HIGHLIGHT:      Here, we design, annotate, and release to the research community a new dataset for fine-grained disinformation analysis that (i) focuses on COVID-19, (ii) combines the perspectives and the interests of journalists, fact-checkers, social media platforms, policy makers, and society as a whole, and (iii) covers both English and Arabic.

84, TITLE:      Unsupervised Transfer of Semantic Role Models from Verbal to Nominal Domain
http://arxiv.org/abs/2005.00278
AUTHORS:        Yanpeng Zhao ; Ivan Titov
HIGHLIGHT:      In this work, we investigate a transfer scenario where we assume role-annotated data for the source verbal domain but only unlabeled data for the target nominal domain.

85, TITLE:      Generating Persona-Consistent Dialogue Responses Using Deep Reinforcement Learning
http://arxiv.org/abs/2005.00036
AUTHORS:        Mohsen Mesgar ; Edwin Simpson ; Yue Wang ; Iryna Gurevych
HIGHLIGHT:      We propose a novel approach to train transformer-based dialogue agents using actor-critic reinforcement learning.

86, TITLE:      Progressively Pretrained Dense Corpus Index for Open-Domain Question Answering
http://arxiv.org/abs/2005.00038
AUTHORS:        Wenhan Xiong ; Hong Wang ; William Yang Wang
HIGHLIGHT:      In this work, we propose a simple and resource-efficient method to pretrain the paragraph encoder.

87, TITLE:      Method for Customizable Automated Tagging: Addressing the Problem of Over-tagging and Under-tagging Text Documents
http://arxiv.org/abs/2005.00042
AUTHORS:        Maharshi R. Pandya ; Jessica Reyes ; Bob Vanderheyden
COMMENTS:       Work done by Maharshi R. Pandya and Jessica Reyes as IBM interns under leadership of Bob Vanderheyden. Article to be published
HIGHLIGHT:      In this paper, we present a method to generate a universal set of tags that can be applied widely to a large document corpus.

88, TITLE:      Facilitating Access to Multilingual COVID-19 Information via Neural Machine Translation
http://arxiv.org/abs/2005.00283
AUTHORS:        Andy Way ; Rejwanul Haque ; Guodong Xie ; Federico Gaspari ; Maja Popovic ; Alberto Poncelas
HIGHLIGHT:      Facilitating Access to Multilingual COVID-19 Information via Neural Machine Translation

89, TITLE:      Multi-Camera Trajectory Forecasting: Pedestrian Trajectory Prediction in a Network of Cameras
http://arxiv.org/abs/2005.00282
AUTHORS:        Olly Styles ; Tanaya Guha ; Victor Sanchez ; Alex Kot
COMMENTS:       CVPR 2020 Precognition workshop
HIGHLIGHT:      We introduce the task of multi-camera trajectory forecasting (MCTF), where the future trajectory of an object is predicted in a network of cameras. To facilitate research in this new area, we release the Warwick-NTU Multi-camera Forecasting Database (WNMF), a unique dataset of multi-camera pedestrian trajectories from a network of 15 synchronized cameras.

90, TITLE:      Distilling Spikes: Knowledge Distillation in Spiking Neural Networks
http://arxiv.org/abs/2005.00288
AUTHORS:        Ravi Kumar Kushawaha ; Saurabh Kumar ; Biplab Banerjee ; Rajbabu Velmurugan
COMMENTS:       Preprint: Manuscript under review
HIGHLIGHT:      In this paper, we propose techniques for knowledge distillation in spiking neural networks for the task of image classification.

91, TITLE:      Context based Text-generation using LSTM networks
http://arxiv.org/abs/2005.00048
AUTHORS:        Sivasurya Santhanam
COMMENTS:       10 pages, Abstract published in A2IC 2018 (https://www.premc.org/doc/A2IC2018/A2IC2018_Book_Of_Abstracts.pdf)
HIGHLIGHT:      Several methods of extracting the context vectors are proposed in this work.

92, TITLE:      Hitachi at SemEval-2020 Task 12: Offensive Language Identification with Noisy Labels using Statistical Sampling and Post-Processing
http://arxiv.org/abs/2005.00295
AUTHORS:        Manikandan Ravikiran ; Amin Ekant Muljibhai ; Toshinori Miyoshi ; Hiroaki Ozaki ; Yuta Koreeda ; Sakata Masayuki
COMMENTS:       preprint v1, Under submission for SemEval 2020 Workshop
HIGHLIGHT:      In this paper, we present our participation in SemEval-2020 Task-12 Subtask-A (English Language) which focuses on offensive language identification from noisy labels.

93, TITLE:      UiO-UvA at SemEval-2020 Task 1: Contextualised Embeddings for Lexical Semantic Change Detection
http://arxiv.org/abs/2005.00050
AUTHORS:        Andrey Kutuzov ; Mario Giulianelli
HIGHLIGHT:      We apply contextualised word embeddings to lexical semantic change detection in the SemEval-2020 Shared Task 1.

94, TITLE:      MAD-X: An Adapter-based Framework for Multi-task Cross-lingual Transfer
http://arxiv.org/abs/2005.00052
AUTHORS:        Jonas Pfeiffer ; Ivan Vulić ; Iryna Gurevych ; Sebastian Ruder
HIGHLIGHT:      We propose MAD-X, an adapter-based framework that enables high portability and parameter-efficient transfer to arbitrary tasks and languages by learning modular language and task representations.

95, TITLE:      CP-NAS: Child-Parent Neural Architecture Search for 1-bit CNNs
http://arxiv.org/abs/2005.00057
AUTHORS:        Li'an Zhuo ; Baochang Zhang ; Hanlin Chen ; Linlin Yang ; Chen Chen ; Yanjun Zhu ; David Doermann
COMMENTS:       7 pages, 6 figures
HIGHLIGHT:      To this end, a Child-Parent (CP) model is introduced to a differentiable NAS to search the binarized architecture (Child) under the supervision of a full-precision model (Parent).

96, TITLE:      Bridging Mode Connectivity in Loss Landscapes and Adversarial Robustness
http://arxiv.org/abs/2005.00060
AUTHORS:        Pu Zhao ; Pin-Yu Chen ; Payel Das ; Karthikeyan Natesan Ramamurthy ; Xue Lin
COMMENTS:       accepted by ICLR 2020
HIGHLIGHT:      In this work, we propose to employ mode connectivity in loss landscapes to study the adversarial robustness of deep neural networks, and provide novel methods for improving this robustness.

97, TITLE:      Attribution Analysis of Grammatical Dependencies in LSTMs
http://arxiv.org/abs/2005.00062
AUTHORS:        Yiding Hao
HIGHLIGHT:      LSTM language models have been shown to capture syntax-sensitive grammatical dependencies such as subject-verb agreement with a high degree of accuracy (Linzen et al., 2016, inter alia).

98, TITLE:      Occlusion resistant learning of intuitive physics from videos
http://arxiv.org/abs/2005.00069
AUTHORS:        Ronan Riochet ; Josef Sivic ; Ivan Laptev ; Emmanuel Dupoux
HIGHLIGHT:      In this work we propose a probabilistic formulation of learning intuitive physics in 3D scenes with significant inter-object occlusions.

99, TITLE:      Importance Driven Continual Learning for Segmentation Across Domains
http://arxiv.org/abs/2005.00079
AUTHORS:        Sinan Özgür Özgün ; Anne-Marie Rickmann ; Abhijit Guha Roy ; Christian Wachinger
HIGHLIGHT:      In this work, we propose a Continual Learning approach for brain segmentation, where a single network is consecutively trained on samples from different domains.

100, TITLE:     Revisiting Unsupervised Relation Extraction
http://arxiv.org/abs/2005.00087
AUTHORS:        Thy Thy Tran ; Phong Le ; Sophia Ananiadou
COMMENTS:       8 pages, 1 figure, 2 tables. Accepted in ACL 2020
HIGHLIGHT:      We conduct a comparison and evaluation of our findings with other URE techniques, to ascertain the important features in URE.

101, TITLE:     Domain Siamese CNNs for Sparse Multispectral Disparity Estimation
http://arxiv.org/abs/2005.00088
AUTHORS:        David-Alexandre Beaupre ; Guillaume-Alexandre Bilodeau
HIGHLIGHT:      In this paper, we propose a new CNN architecture able to do disparity estimation between images from different spectrum, namely thermal and visible in our case.

102, TITLE:     AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages
http://arxiv.org/abs/2005.00085
AUTHORS:        Anoop Kunchukuttan ; Divyanshu Kakwani ; Satish Golla ; Gokul N. C. ; Avik Bhattacharyya ; Mitesh M. Khapra ; Pratyush Kumar
COMMENTS:       7 pages, 8 tables, https://github.com/ai4bharat-indicnlp/indicnlp_corpus
HIGHLIGHT:      We present the IndicNLP corpus, a large-scale, general-domain corpus containing 2.7 billion words for 10 Indian languages from two language families. We create news article category classification datasets for 9 languages to evaluate the embeddings.

103, TITLE:     Aspect-Controlled Neural Argument Generation
http://arxiv.org/abs/2005.00084
AUTHORS:        Benjamin Schiller ; Johannes Daxenberger ; Iryna Gurevych
HIGHLIGHT:      In this work, we train a language model for argument generation that can be controlled on a fine-grained level to generate sentence-level arguments for a given topic, stance, and aspect.

104, TITLE:     An Early Study on Intelligent Analysis of Speech under COVID-19: Severity, Sleep Quality, Fatigue, and Anxiety
http://arxiv.org/abs/2005.00096
AUTHORS:        Jing Han ; Kun Qian ; Meishu Song ; Zijiang Yang ; Zhao Ren ; Shuo Liu ; Juan Liu ; Huaiyuan Zheng ; Wei Ji ; Tomoya Koike ; Xiao Li ; Zixing Zhang ; Yoshiharu Yamamoto ; Björn W. Schuller
HIGHLIGHT:      In this study, we focus on developing some potential use-cases of intelligent speech analysis for COVID-19 diagnosed patients.



==========Updates to Previous Papers==========
1, TITLE:       Improving Vision-and-Language Navigation with Image-Text Pairs from the Web
http://arxiv.org/abs/2004.14973
AUTHORS:        Arjun Majumdar ; Ayush Shrivastava ; Stefan Lee ; Peter Anderson ; Devi Parikh ; Dhruv Batra
HIGHLIGHT:      Specifically, we develop VLN-BERT, a visiolinguistic transformer-based model for scoring the compatibility between an instruction ('...stop at the brown sofa') and a sequence of panoramic RGB images captured by the agent.

2, TITLE:       Analyzing Political Parody in Social Media
http://arxiv.org/abs/2004.13878
AUTHORS:        Antonis Maronikolakis ; Danae Sanchez Villegas ; Daniel Preotiuc-Pietro ; Nikolaos Aletras
HIGHLIGHT:      In this paper, we present the first computational study of parody. We introduce a new publicly available data set of tweets from real politicians and their corresponding parody accounts.

3, TITLE:       An initial attempt of combining visual selective attention with deep reinforcement learning
http://arxiv.org/abs/1811.04407
AUTHORS:        Liu Yuezhang ; Ruohan Zhang ; Dana H. Ballard
COMMENTS:       7 pages, 8 figures, rejected by AAAI 2019 Workshop on Reinforcement Learning and Games
HIGHLIGHT:      We visualize and analyze the feature maps of DQN on a toy problem Catch, and propose an approach to combine visual selective attention with deep reinforcement learning.

4, TITLE:       Optimising Lockdown Policies for Epidemic Control using Reinforcement Learning
http://arxiv.org/abs/2003.14093
AUTHORS:        Harshad Khadilkar ; Tanuja Ganu ; Deva P Seetharam
HIGHLIGHT:      In this working paper, we present a quantitative way to compute lockdown decisions for individual cities or regions, while balancing health and economic considerations.

5, TITLE:       Posterior Calibrated Training on Sentence Classification Tasks
http://arxiv.org/abs/2004.14500
AUTHORS:        Taehee Jung ; Dongyeop Kang ; Hua Cheng ; Lucas Mentch ; Thomas Schaaf
COMMENTS:       Accepted at ACL 2020
HIGHLIGHT:      Here we propose an end-to-end training procedure called posterior calibrated (PosCal) training that directly optimizes the objective while minimizing the difference between the predicted and empirical posterior probabilities.We show that PosCal not only helps reduce the calibration error but also improve task performance by penalizing drops in performance of both objectives.

6, TITLE:       Event-based Robotic Grasping Detection with Neuromorphic Vision Sensor and Event-Stream Dataset
http://arxiv.org/abs/2004.13652
AUTHORS:        Bin Li ; Hu Cao ; Zhongnan Qu ; Yingbai Hu ; Zhenke Wang ; Zichen Liang
COMMENTS:       submit to the Frontiers Neurorobotics
HIGHLIGHT:      In this work, we consider the problem of detecting robotic grasps in a moving camera view of a scene containing objects. We construct a robotic grasping dataset named Event-Stream Dataset with 91 objects. This work provides a large-scale and well-annotated dataset, and promotes the neuromorphic vision applications in agile robot.

7, TITLE:       Temporally Coherent Embeddings for Self-Supervised Video Representation Learning
http://arxiv.org/abs/2004.02753
AUTHORS:        Joshua Knights ; Anthony Vanderkop ; Daniel Ward ; Olivia Mackenzie-Ross ; Peyman Moghadam
COMMENTS:       Under review! Project page: https://csiro-robotics.github.io/TCE_Webpage/
HIGHLIGHT:      This paper presents TCE: Temporally Coherent Embeddings for self-supervised video representation learning.

8, TITLE:       Hierarchically Fair Federated Learning
http://arxiv.org/abs/2004.10386
AUTHORS:        Jingfeng Zhang ; Cheng Li ; Antonio Robles-Kelly ; Mohan Kankanhalli
HIGHLIGHT:      We propose a novel hierarchically fair federated learning (HFFL) framework.

9, TITLE:       Fact or Fiction: Verifying Scientific Claims
http://arxiv.org/abs/2004.14974
AUTHORS:        David Wadden ; Kyle Lo ; Lucy Lu Wang ; Shanchuan Lin ; Madeleine van Zuylen ; Arman Cohan ; Hannaneh Hajishirzi
COMMENTS:       16 pages (including appendices), 10 figures, 7 tables. GitHub: https://github.com/allenai/scifact
HIGHLIGHT:      We present a baseline model and assess its performance on SciFact.

10, TITLE:      Conditional Augmentation for Aspect Term Extraction via Masked Sequence-to-Sequence Generation
http://arxiv.org/abs/2004.14769
AUTHORS:        Kun Li ; Chengbo Chen ; Xiaojun Quan ; Qing Ling ; Yan Song
COMMENTS:       To appear at ACL 2020
HIGHLIGHT:      In this paper, we formulate the data augmentation as a conditional generation task: generating a new sentence while preserving the original opinion targets and labels.

11, TITLE:      Angle-based Search Space Shrinking for Neural Architecture Search
http://arxiv.org/abs/2004.13431
AUTHORS:        Yiming Hu ; Yuding Liang ; Zichao Guo ; Ruosi Wan ; Xiangyu Zhang ; Yichen Wei ; Qingyi Gu ; Jian Sun
COMMENTS:       15 pages
HIGHLIGHT:      In this work, we present a simple and general search space shrinking method, called Angle-Based search space Shrinking (ABS), for Neural Architecture Search (NAS).

12, TITLE:      Self-Supervised and Controlled Multi-Document Opinion Summarization
http://arxiv.org/abs/2004.14754
AUTHORS:        Hady Elsahar ; Maximin Coavoux ; Matthias Gallé ; Jos Rozen
COMMENTS:       18 pages including 5 pages appendix
HIGHLIGHT:      We propose a self-supervised setup that considers an individual document as a target summary for a set of similar documents.

13, TITLE:      FastSurfer -- A fast and accurate deep learning based neuroimaging pipeline
http://arxiv.org/abs/1910.03866
AUTHORS:        Leonie Henschel ; Sailesh Conjeti ; Santiago Estrada ; Kersten Diers ; Bruce Fischl ; Martin Reuter
COMMENTS:       Submitted to NeuroImage
HIGHLIGHT:      In this work we propose a fast and accurate deep learning based neuroimaging pipeline for the automated processing of structural human brain MRI scans, replicating FreeSurfer's anatomical segmentation including surface reconstruction and cortical parcellation.

14, TITLE:      Unsatisfiability Proofs for Weight 16 Codewords in Lam's Problem
http://arxiv.org/abs/2001.11973
AUTHORS:        Curtis Bright ; Kevin K. H. Cheung ; Brett Stevens ; Ilias Kotsireas ; Vijay Ganesh
COMMENTS:       To appear in Proceedings of the 29th International Joint Conference on Artificial Intelligence (IJCAI 2020)
HIGHLIGHT:      We performed a verification of these searches by reducing the problem to the Boolean satisfiability problem (SAT).

15, TITLE:      Meta Answering for Machine Reading
http://arxiv.org/abs/1911.04156
AUTHORS:        Benjamin Borschinger ; Jordan Boyd-Graber ; Christian Buck ; Jannis Bulian ; Massimiliano Ciaramita ; Michelle Chen Huebscher ; Wojciech Gajewski ; Yannic Kilcher ; Rodrigo Nogueira ; Lierni Sestorain Saralegu
HIGHLIGHT:      We investigate a framework for machine reading, inspired by real world information-seeking problems, where a meta question answering system interacts with a black box environment.

16, TITLE:      On the Possibilities and Limitations of Multi-hop Reasoning Under Linguistic Imperfections
http://arxiv.org/abs/1901.02522
AUTHORS:        Daniel Khashabi ; Erfan Sadeqi Azer ; Tushar Khot ; Ashish Sabharwal ; Dan Roth
HIGHLIGHT:      We present the first formal framework to study such empirical observations.

17, TITLE:      Perturbations on the Perceptual Ball
http://arxiv.org/abs/1912.09405
AUTHORS:        Andrew Elliott ; Stephen Law ; Chris Russell
COMMENTS:       First two authors contributed equally to this work. Preprint under review
HIGHLIGHT:      We present a simple regularisation of Adversarial Perturbations based upon the perceptual loss.

18, TITLE:      An Unsupervised Joint System for Text Generation from Knowledge Graphs and Semantic Parsing
http://arxiv.org/abs/1904.09447
AUTHORS:        Martin Schmitt ; Sahand Sharifzadeh ; Volker Tresp ; Hinrich Schütze
HIGHLIGHT:      To this end, we present the first approach to fully unsupervised text generation from KGs and KG generation from text.

19, TITLE:      Unifying graded and parameterised monads
http://arxiv.org/abs/2001.10274
AUTHORS:        Dominic Orchard ; Philip Wadler ; Harley Eades III
COMMENTS:       In Proceedings MSFP 2020, arXiv:2004.14735
HIGHLIGHT:      Using this as a basis, we show how graded and parameterised monads can be unified, studying their similarities and differences along the way.

20, TITLE:      Adversarial Augmentation Policy Search for Domain and Cross-Lingual Generalization in Reading Comprehension
http://arxiv.org/abs/2004.06076
AUTHORS:        Adyasha Maharana ; Mohit Bansal
COMMENTS:       15 pages (v2: added Turkish in addition to Russian and German for cross-lingual experiments)
HIGHLIGHT:      We address this issue via RL and more efficient Bayesian policy search methods for automatically learning the best augmentation policy combinations of the transformation probability for each adversary in a large search space.

21, TITLE:      MGCN: Descriptor Learning using Multiscale GCNs
http://arxiv.org/abs/2001.10472
AUTHORS:        Yiqun Wang ; Jing Ren ; Dong-Ming Yan ; Jianwei Guo ; Xiaopeng Zhang ; Peter Wonka
COMMENTS:       Accepted to SIGGRAPH 2020. (15 pages, 15 figures, 12 tables, low-resolution version)
HIGHLIGHT:      We propose a novel framework for computing descriptors for characterizing points on three-dimensional surfaces.

22, TITLE:      A Span-based Linearization for Constituent Trees
http://arxiv.org/abs/2004.14704
AUTHORS:        Yang Wei ; Yuanbin Wu ; Man Lan
COMMENTS:       Accepted to ACL 2020
HIGHLIGHT:      We propose a novel linearization of a constituent tree, together with a new locally normalized model.

23, TITLE:      TXtract: Taxonomy-Aware Knowledge Extraction for Thousands of Product Categories
http://arxiv.org/abs/2004.13852
AUTHORS:        Giannis Karamanolakis ; Jun Ma ; Xin Luna Dong
COMMENTS:       Accepted to ACL 2020 (Long Paper)
HIGHLIGHT:      This paper proposes TXtract, a taxonomy-aware knowledge extraction model that applies to thousands of product categories organized in a hierarchical taxonomy.

24, TITLE:      Opportunities and Challenges of Deep Learning Methods for Electrocardiogram Data: A Systematic Review
http://arxiv.org/abs/2001.01550
AUTHORS:        Shenda Hong ; Yuxi Zhou ; Junyuan Shang ; Cao Xiao ; Jimeng Sun
COMMENTS:       Accepted by Computers in Biology and Medicine
HIGHLIGHT:      Objective:This paper presents a systematic review of deep learning methods for ECG data from both modeling and application perspectives.

25, TITLE:      Know Your Surroundings: Exploiting Scene Information for Object Tracking
http://arxiv.org/abs/2003.11014
AUTHORS:        Goutam Bhat ; Martin Danelljan ; Luc Van Gool ; Radu Timofte
HIGHLIGHT:      In this work, we propose a novel tracking architecture which can utilize scene information for tracking.

26, TITLE:      Real-time, Universal, and Robust Adversarial Attacks Against Speaker Recognition Systems
http://arxiv.org/abs/2003.02301
AUTHORS:        Yi Xie ; Cong Shi ; Zhuohang Li ; Jian Liu ; Yingying Chen ; Bo Yuan
COMMENTS:       Published as a conference paper at ICASSP 2020
HIGHLIGHT:      In this paper, we propose the first real-time, universal, and robust adversarial attack against the state-of-the-art deep neural network (DNN) based speaker recognition system.

27, TITLE:      Sentence Meta-Embeddings for Unsupervised Semantic Textual Similarity
http://arxiv.org/abs/1911.03700
AUTHORS:        Nina Poerner ; Ulli Waltinger ; Hinrich Schütze
HIGHLIGHT:      We apply, extend and evaluate different meta-embedding methods from the word embedding literature at the sentence level, including dimensionality reduction (Yin and Sch\"utze, 2016), generalized Canonical Correlation Analysis (Rastogi et al., 2015) and cross-view auto-encoders (Bollegala and Bao, 2018).

28, TITLE:      Graph Convolutions over Constituent Trees for Syntax-Aware Semantic Role Labeling
http://arxiv.org/abs/1909.09814
AUTHORS:        Diego Marcheggiani ; Ivan Titov
HIGHLIGHT:      In contrast, we show how graph convolutional networks (GCNs) can be used to encode constituent structures and inform an SRL system.

29, TITLE:      Fine-grained hardness of CVP(P) -- Everything that we can prove (and nothing else)
http://arxiv.org/abs/1911.02440
AUTHORS:        Divesh Aggarwal ; Huck Bennett ; Alexander Golovnev ; Noah Stephens-Davidowitz
HIGHLIGHT:      We show a number of fine-grained hardness results for the Closest Vector Problem in the $\ell_p$ norm ($\mathrm{CVP}_p$), and its approximate and non-uniform variants.

30, TITLE:      Contextualized Sparse Representations for Real-Time Open-Domain Question Answering
http://arxiv.org/abs/1911.02896
AUTHORS:        Jinhyuk Lee ; Minjoon Seo ; Hannaneh Hajishirzi ; Jaewoo Kang
COMMENTS:       ACL 2020
HIGHLIGHT:      In this paper, we aim to improve the quality of each phrase embedding by augmenting it with a contextualized sparse representation (Sparc).

31, TITLE:      Unsupervised Domain Clusters in Pretrained Language Models
http://arxiv.org/abs/2004.02105
AUTHORS:        Roee Aharoni ; Yoav Goldberg
COMMENTS:       Accepted as a long paper in ACL 2020
HIGHLIGHT:      We harness this property and propose domain data selection methods based on such models, which require only a small set of in-domain monolingual data.

32, TITLE:      FSPool: Learning Set Representations with Featurewise Sort Pooling
http://arxiv.org/abs/1906.02795
AUTHORS:        Yan Zhang ; Jonathon Hare ; Adam Prügel-Bennett
COMMENTS:       Published at International Conference on Learning Representations (ICLR) 2020
HIGHLIGHT:      We introduce a pooling method for sets of feature vectors based on sorting features across elements of the set.

33, TITLE:      Robust Cross-lingual Embeddings from Parallel Sentences
http://arxiv.org/abs/1912.12481
AUTHORS:        Ali Sabet ; Prakhar Gupta ; Jean-Baptiste Cordonnier ; Robert West ; Martin Jaggi
HIGHLIGHT:      We propose a bilingual extension of the CBOW method which leverages sentence-aligned corpora to obtain robust cross-lingual word and sentence representations.

34, TITLE:      Multilingual Embeddings Jointly Induced from Contexts and Concepts: Simple, Strong and Scalable
http://arxiv.org/abs/1811.00586
AUTHORS:        Philipp Dufter ; Mengjie Zhao ; Hinrich Schütze
HIGHLIGHT:      In this paper, we propose Co+Co, a simple and scalable method that combines context-based and concept-based learning.

35, TITLE:      Improved Natural Language Generation via Loss Truncation
http://arxiv.org/abs/2004.14589
AUTHORS:        Daniel Kang ; Tatsunori Hashimoto
COMMENTS:       ACL 2020 Camera Ready Submission
HIGHLIGHT:      In this work, we show that the distinguishability of the models and reference serves as a principled and robust alternative for handling invalid references.

36, TITLE:      Inf-Net: Automatic COVID-19 Lung Infection Segmentation from CT Scans
http://arxiv.org/abs/2004.14133
AUTHORS:        Deng-Ping Fan ; Tao Zhou ; Ge-Peng Ji ; Yi Zhou ; Geng Chen ; Huazhu Fu ; Jianbing Shen ; Ling Shao
COMMENTS:       The project page can be found in: https://github.com/DengPingFan/Inf-Net
HIGHLIGHT:      Our semi-supervised framework can improve the learning ability and achieve a higher performance.

37, TITLE:      CNN Explainer: Learning Convolutional Neural Networks with Interactive Visualization
http://arxiv.org/abs/2004.15004
AUTHORS:        Zijie J. Wang ; Robert Turko ; Omar Shaikh ; Haekyu Park ; Nilaksh Das ; Fred Hohman ; Minsuk Kahng ; Duen Horng Chau
COMMENTS:       11 pages, 14 figures. For a demo video, see https://youtu.be/HnWIHWFbuUQ For a live demo, visit https://poloclub.github.io/cnn-explainer/
HIGHLIGHT:      We present CNN Explainer, an interactive visualization tool designed for non-experts to learn and examine convolutional neural networks (CNNs), a foundational deep learning model architecture.

38, TITLE:      MKD: a Multi-Task Knowledge Distillation Approach for Pretrained Language Models
http://arxiv.org/abs/1911.03588
AUTHORS:        Linqing Liu ; Huan Wang ; Jimmy Lin ; Richard Socher ; Caiming Xiong
HIGHLIGHT:      In this paper, we explore knowledge distillation under the multi-task learning setting.

39, TITLE:      Two Birds, One Stone: A Simple, Unified Model for Text Generation from Structured and Unstructured Data
http://arxiv.org/abs/1909.10158
AUTHORS:        Hamidreza Shahidi ; Ming Li ; Jimmy Lin
COMMENTS:       Accepted at ACL 2020
HIGHLIGHT:      In this work, we show that this is also the case for text generation from structured and unstructured data.

40, TITLE:      Universal Dependencies according to BERT: both more specific and more general
http://arxiv.org/abs/2004.14620
AUTHORS:        Tomasz Limisiewicz ; Rudolf Rosa ; David Mareček
HIGHLIGHT:      We suggest a method for relation identification and syntactic tree construction.

41, TITLE:      Alternative Function Approximation Parameterizations for Solving Games: An Analysis of $f$-Regression Counterfactual Regret Minimization
http://arxiv.org/abs/1912.02967
AUTHORS:        Ryan D'Orazio ; Dustin Morrill ; James R. Wright ; Michael Bowling
COMMENTS:       11 pages, includes appendix
HIGHLIGHT:      We derive approximation error-aware regret bounds for $(\Phi, f)$-regret matching, which applies to a general class of link functions and regret objectives.

42, TITLE:      On the Degree of Boolean Functions as Polynomials over $\mathbb{Z}_m$
http://arxiv.org/abs/1910.12458
AUTHORS:        Xiaoming Sun ; Yuan Sun ; Jiaheng Wang ; Kewen Wu ; Zhiyu Xia ; Yufan Zheng
COMMENTS:       To appear in ICALP'20
HIGHLIGHT:      In this paper, we investigate the lower bound of modulo-$m$ degree of Boolean functions.

43, TITLE:      NUBIA: NeUral Based Interchangeability Assessor for Text Generation
http://arxiv.org/abs/2004.14667
AUTHORS:        Hassan Kane ; Muhammed Yusuf Kocyigit ; Ali Abdalla ; Pelkins Ajanoh ; Mohamed Coulibali
COMMENTS:       8 pages, 5 tables, and 2 figures
HIGHLIGHT:      We present NUBIA, a methodology to build automatic evaluation metrics for text generation using only machine learning models as core components.

44, TITLE:      Adversarial Color Enhancement: Generating Unrestricted Adversarial Images by Optimizing a Color Filter
http://arxiv.org/abs/2002.01008
AUTHORS:        Zhengyu Zhao ; Zhuoran Liu ; Martha Larson
COMMENTS:       Code available at https://github.com/ZhengyuZhao/ACE
HIGHLIGHT:      We introduce an approach that enhances images using a color filter in order to create adversarial effects, which fool neural networks into misclassification.

45, TITLE:      Dialogue Transformers
http://arxiv.org/abs/1910.00486
AUTHORS:        Vladimir Vlasov ; Johannes E. M. Mosig ; Alan Nichol
COMMENTS:       10 pages, 4 figures, 1 table
HIGHLIGHT:      We introduce a dialogue policy based on a transformer architecture, where the self-attention mechanism operates over the sequence of dialogue turns.

46, TITLE:      Learning What to Learn for Video Object Segmentation
http://arxiv.org/abs/2003.11540
AUTHORS:        Goutam Bhat ; Felix Järemo Lawin ; Martin Danelljan ; Andreas Robinson ; Michael Felsberg ; Luc Van Gool ; Radu Timofte
COMMENTS:       First two authors contributed equally
HIGHLIGHT:      We address this by introducing an end-to-end trainable VOS architecture that integrates a differentiable few-shot learning module.

47, TITLE:      Context is Key: Grammatical Error Detection with Contextual Word Representations
http://arxiv.org/abs/1906.06593
AUTHORS:        Samuel Bell ; Helen Yannakoudakis ; Marek Rei
HIGHLIGHT:      In this paper, we perform a systematic comparison of ELMo, BERT and Flair embeddings (Peters et al., 2017; Devlin et al., 2018; Akbik et al., 2018) on a range of public GED datasets, and propose an approach to effectively integrate such representations in current methods, achieving a new state of the art on GED.

48, TITLE:      Evaluating Transformer-Based Multilingual Text Classification
http://arxiv.org/abs/2004.13939
AUTHORS:        Sophie Groenwold ; Samhita Honnavalli ; Lily Ou ; Aesha Parekh ; Sharon Levy ; Diba Mirza ; William Yang Wang
COMMENTS:       Total of 15 pages (9 pages for paper, 2 pages for references, 4 pages for appendix). Changed title
HIGHLIGHT:      Through a detailed discussion of word order typology, morphological typology, and comparative linguistics, we identify which variables most affect language modeling efficacy; in addition, we calculate word order and morphological similarity indices to aid our empirical study.

49, TITLE:      Hooks in the Headline: Learning to Generate Headlines with Controlled Styles
http://arxiv.org/abs/2004.01980
AUTHORS:        Di Jin ; Zhijing Jin ; Joey Tianyi Zhou ; Lisa Orii ; Peter Szolovits
COMMENTS:       ACL 2020
HIGHLIGHT:      We propose a new task, Stylistic Headline Generation (SHG), to enrich the headlines with three style options (humor, romance and clickbait), in order to attract more readers.

50, TITLE:      Zero-shot Entity Linking with Dense Entity Retrieval
http://arxiv.org/abs/1911.03814
AUTHORS:        Ledell Wu ; Fabio Petroni ; Martin Josifoski ; Sebastian Riedel ; Luke Zettlemoyer
HIGHLIGHT:      In this paper, we introduce a simple and effective two-stage approach for zero-shot linking, based on fine-tuned BERT architectures.

51, TITLE:      Adversarial Language Games for Advanced Natural Language Intelligence
http://arxiv.org/abs/1911.01622
AUTHORS:        Yuan Yao ; Haoxi Zhong ; Zhengyan Zhang ; Xu Han ; Xiaozhi Wang ; Chaojun Xiao ; Guoyang Zeng ; Zhiyuan Liu ; Maosong Sun
COMMENTS:       Work in progress
HIGHLIGHT:      In this work, we propose a challenging adversarial language game called Adversarial Taboo as an example, in which an attacker and a defender compete around a target word.

52, TITLE:      CCMatrix: Mining Billions of High-Quality Parallel Sentences on the WEB
http://arxiv.org/abs/1911.04944
AUTHORS:        Holger Schwenk ; Guillaume Wenzek ; Sergey Edunov ; Edouard Grave ; Armand Joulin
COMMENTS:       13 pages, 4 figures. arXiv admin note: text overlap with arXiv:1907.05791
HIGHLIGHT:      We show that margin-based bitext mining in a multilingual sentence space can be applied to monolingual corpora of billions of sentences.

53, TITLE:      Weakly Supervised Dataset Collection for Robust Person Detection
http://arxiv.org/abs/2003.12263
AUTHORS:        Munetaka Minoguchi ; Ken Okayama ; Yutaka Satoh ; Hirokatsu Kataoka
COMMENTS:       Project page: https://github.com/cvpaperchallenge/FashionCultureDataBase_DLoader The paper is under consideration at Pattern Recognition Letters
HIGHLIGHT:      To construct an algorithm that can provide robust person detection, we present a dataset with over 8 million images that was produced in a weakly supervised manner.

54, TITLE:      Don't Stop Pretraining: Adapt Language Models to Domains and Tasks
http://arxiv.org/abs/2004.10964
AUTHORS:        Suchin Gururangan ; Ana Marasović ; Swabha Swayamdipta ; Kyle Lo ; Iz Beltagy ; Doug Downey ; Noah A. Smith
COMMENTS:       ACL 2020
HIGHLIGHT:      We present a study across four domains (biomedical and computer science publications, news, and reviews) and eight classification tasks, showing that a second phase of pretraining in-domain (domain-adaptive pretraining) leads to performance gains, under both high- and low-resource settings.

55, TITLE:      CamemBERT: a Tasty French Language Model
http://arxiv.org/abs/1911.03894
AUTHORS:        Louis Martin ; Benjamin Muller ; Pedro Javier Ortiz Suárez ; Yoann Dupont ; Laurent Romary ; Éric Villemonte de la Clergerie ; Djamé Seddah ; Benoît Sagot
COMMENTS:       ACL 2020 long paper. Web site: https://camembert-model.fr
HIGHLIGHT:      In this paper, we investigate the feasibility of training monolingual Transformer-based language models for other languages, taking French as an example and evaluating our language models on part-of-speech tagging, dependency parsing, named entity recognition and natural language inference tasks.

56, TITLE:      End-to-end Recurrent Multi-Object Tracking and Trajectory Prediction with Relational Reasoning
http://arxiv.org/abs/1907.12887
AUTHORS:        Fabian B. Fuchs ; Adam R. Kosiorek ; Li Sun ; Oiwi Parker Jones ; Ingmar Posner
HIGHLIGHT:      Building upon HART, a neural class-agnostic single-object tracker, we introduce a multi-object tracking method MOHART capable of relational reasoning.

57, TITLE:      A Popperian Falsification of Artificial Intelligence -- Lighthill Defended
http://arxiv.org/abs/1704.08111
AUTHORS:        Steven Meyer
COMMENTS:       12 pages. Version improves discussion of chess and adds sections on when combinatorial explosion may not apply
HIGHLIGHT:      The paper describes the Popperian method and discusses Paul Nurse's application of the method to cell biology that also involves questions of mechanism and behavior.

58, TITLE:      Winning an Election: On Emergent Strategic Communication in Multi-Agent Networks
http://arxiv.org/abs/1902.06897
AUTHORS:        Shubham Gupta ; Ambedkar Dukkipati
COMMENTS:       A shorter version of this paper has been accepted as an extended abstract at AAMAS 2020
HIGHLIGHT:      We formulate the problem using a voting game where two candidate agents contest in an election with the goal of convincing population members (other agents), that are connected to each other via an underlying network, to vote for them.

59, TITLE:      E-BERT: Efficient-Yet-Effective Entity Embeddings for BERT
http://arxiv.org/abs/1911.03681
AUTHORS:        Nina Poerner ; Ulli Waltinger ; Hinrich Schütze
HIGHLIGHT:      We present a novel way of injecting factual knowledge about entities into the pretrained BERT model (Devlin et al., 2019): We align Wikipedia2Vec entity vectors (Yamada et al., 2016) with BERT's native wordpiece vector space and use the aligned entity vectors as if they were wordpiece vectors.

60, TITLE:      Inexpensive Domain Adaptation of Pretrained Language Models: Case Studies on Biomedical NER and Covid-19 QA
http://arxiv.org/abs/2004.03354
AUTHORS:        Nina Poerner ; Ulli Waltinger ; Hinrich Schütze
HIGHLIGHT:      We cover over 50% of the BioBERT-BERT F1 delta, at 5% of BioBERT's CO_2 footprint and 2% of its cloud compute cost.