Guideline:

• paper reading讲解的时候要深入浅出，确保自己看懂了，再用通俗的话讲出来。关键是把文章工作讲清楚，motivation，方法部分，实验是否支撑，该工作的优点和缺点，对你个人工作的启发。每部分大概2~3页slides即可。最重要的是后面两部分，需要你自己对工作批判性的阅读。
• 时间暂定是周4下午。 如果人不齐的话提前告知，视情况再确定时间。
• 分享的同学务必提前告知大家分享的论文，并在分享前update paper信息及slides到 52paper.github.io；新人权限开通请联系jamgao。
• 参与者希望都能够提前把分享的paper进行相关背景的了解，积极提出问题及参与讨论。

Next Meeting

2020/04/23

Speakers	Papers	Slides	Others
kaiwang	-	[slide]	-
-	ICLR 2020 ELECTRA: PRE-TRAINING TEXT ENCODERS AS DISCRIMINATORS RATHER THAN GENERATORS	-	-
-	ACL 2019 Bridging the Gap between Training and Inference for Neural Machine Translation	-	-
-	ACL 2020 Multi-Domain Dialogue Acts and Response Co-Generation	-	-

2020/04/16

Speakers	Papers	Slides	Others
haoyusong	Transfer Learning in Personalized Dialogue Generation	[slide]	-
-	WWW Journal 2019 Neural Personalized Response Generation as Domain Adaptation	-	-
-	AAAI 2019 short TransferTransfo-A Transfer Learning Approach for Neural Network Based Conversational Agents	-	-
-	ACL 2019 short Large-scale transfer learning for natural language generation	-	-
-	AAAI 2020 A Pre-training Based Personalized Dialogue Generation Model with Persona-sparse Data	-	-

2019/12/25

Speakers	Papers	Slides	Others
rickywchen	Dialogue Summarization	[slide]	-
-	ACL2019 (short) Keep Meeting Summaries on Topic: Abstractive Multi-Modal Meeting Summarization	-	-
-	KDD2019 Automatic Dialogue Summary Generation for Customer Service	-	-

2019/09/26

Speakers	Papers	Slides	Others
rickwwang	Some Research Progress on Story Generation	[slide]	-
-	CoNLL2019 Do Massively Pretrained Language Models Make Better Storytellers?	-	-
-	EMNLP2019 Counterfactual Story Reasoning and Generation	-	-

2019/09/19

Speakers	Papers	Slides	Others
hgong	-	[slide]	-
-	AAAI2019 Data-to-Text Generation with Content Selection and Planning	-	-
-	ACL2019 Data-to-text Generation with Entity Modeling	-	-
-	ACL2019 Learning to Select, Track, and Generate for Data-to-Text	-	-

2019/09/10

Speakers	Papers	Slides	Others
jimblin	-	[slide]	-
-	ACL2018 Multi-Turn Response Selection for Chatbots with Deep Attention Matching Network	-	-
-	ACL2019 One Time of Interaction May Not Be Enough: Go Deep with an Interaction-over-Interaction Network for Response Selection in Dialogues	-	-
-	ACL2019 Constructing Interpretive Spatio-Temporal Features for Multi-Turn Response Selection	-	-

2019/08/22

Speakers	Papers	Slides	Others
qintongli	-	[slide]	-
-	AAAI2018 Emotional Chatting Machine: Emotional Conversation Generation with Internal and External Memory	-	-
-	ACL2018 MOJITALK: Generating Emotional Responses at Scale	-	-
-	AAAI2019 An Affect-Rich Neural Conversational Model with Biased Attention and Weighted Cross-Entropy Loss	-	-

2019/08/15

Speakers	Papers	Slides	Others
jiangtongli	ACL Report	[slide]	-
-	ACL2019 Bridging the Gap between Training and Inference for Neural Machine Translation	-	-
-	ACL2019 OpenDialKG: Explainable Conversational Reasoning with Attention-based Walks over Knowledge Graphs	-	-
-	ACL2019 Do Neural Dialog Systems Use the Conversation History Effectively? An Empirical Study	-	-
-	ACL2019 Generating Fluent Adversarial Examples for Natural Languages	-	-
-	ACL2019 Dynamically Fused Graph Network for Multi-hop Reasoning	-	-
-	ACL2019 Multi-step Reasoning via Recurrent Dual Attention for Visual Dialog	-	-

2019/07/25

Speakers	Papers	Slides	Others
jcykcai	-	[slide]	-
-	ACL2019 Analyzing Multi-Head Self-Attention: Specialized Heads Do the Heavy Lifting, the Rest Can Be Pruned	-	-
-	ACL2019 Interpretable Neural Predictions with Differentiable Binary Variables	-	-

2019/07/18

Speakers	Papers	Slides	Others
jiangtongli	Some research progress on sequence generation	[slide]	-
-	Arxiv 2015 How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?	-	-
-	ICML2019 CoT: Cooperative Training for Generative Modeling of Discrete Data	-	-
-	ICLR2019 Improving Sequence-to-Sequence Learning via Optimal Transport	-	-

2019/07/11

Speakers	Papers	Slides	Others
zltian	Triples-to-text generation & its pre-training	[slide]	-
-	INLG2018 Deep Graph Convolutional Encoders for Structured Data to Text Generation	-	-
-	NAACL2019 Step-by-Step: Separating Planning from Realization in Neural Data-to-Text Generation	-	-
-	NIPS(Workshop)2016 Variational Graph Auto-Encoders	-	-

2019/07/04

Speakers	Papers	Slides	Others
rickwwang	Some Research Progress on Story Generation	[slide]	-
-	EMNLP2018 A Skeleton-Based Model for Promoting Coherence Among Sentences in Narrative Story Generation	-	-
-	AAAI2019 Plan-And-Write: Towards Better Automatic Storytelling	-	-
-	ACL2019 Strategies for Structuring Story Generation	-	-

2019/06/28

Speakers	Papers	Slides	Others
jcykcai	Rethinking the generation orders of sequence	[slide]	-
	ICML2019 Insertion Transformer: Flexible Sequence Generation via Insertion Operations	-	-
-	ICML2019 Non-Monotonic Sequential Text Generation	-	-
-	arxiv19 Insertion-based Decoding with automatically Inferred Generation Order	-	-
-	EMNLP18 The Importance of Generation Order in Language Modeling	-	-
-	arxiv19 XLNet: Generalized Autoregressive Pretraining for Language Understanding	-	-

2019/06/17

Speakers	Papers	Slides	Others
gaojun	The Best of Both Worlds: Combining Recent Advances in Neural Machine Translation	-	-
-	How Much Attention Do You Need? A Granular Analysis of Neural Machine Translation Architectures	-	-
-	Why Self-Attention? A Targeted Evaluation of Neural Machine Translation Architectures	-	-
-	Argument Generation with Retrieval, Planning, and Realization	-	-

2019/05/30

Speakers	Papers	Slides	Others
jiachendu	ICLR 2019 LEARNING TO REPRESENT EDITS	[slide]	-
-	Text Infilling	-	-
-	TIGS: An Inference Algorithm for Text Infilling with Gradient Search	-	-

2019/05/09

Speakers	Papers	Slides	Others
lixin	The Curious Case of Neural Text Degeneration	[slide]	-

2019/04/24

Speakers	Papers	Slides	Others
evanyfgao(高一帆)	Reasoning in Multi-hop Reading Comprehension	[slide]	-

2019/04/11

Speakers	Papers	Slides	Others
royrong(荣钰)	Representation Learning on Graphs	[slide]	-

2019/04/04

Speakers	Papers	Slides	Others
jcykcai	AAAI17 Mechanism-Aware Neural Machine for Dialogue Response Generation	[slide]	-
-	ACL18 Unsupervised Discrete Sentence Representation Learning for Interpretable Neural Dialog Generation	-	-
-	EMNLP18 Learning Neural Templates for Text Generation	-	-

2019/03/28

Speakers	Papers	Slides	Others
yxsu	TACL2018 Polite Dialogue Generation Without Parallel Data	[slide]	-

2019/03/21

Speakers	Papers	Slides	Others
gaojun	NIPS2018 Content preserving text generation with attribute controls	[slide]	-
hongyining	EMNLP2017 Challenges in Data-to-Document Generation	[slide]	-
-	Data-to-Text Generation with Content Selection and Planning	-	-

2019/01/18

Speakers	Papers	Slides	Others
zhuqile	ICLR2019 Recent Advances in Autoencoder-Based Representation Learning	[slide]	-
jiangtongli	ICLR2019 Pay Less Attention with Lightweight and Dynamic Convolutions	[slide]	-

2019/01/10

Speakers	Papers	Slides	Others
zhuqile	ICLR2019 Variational Discriminator Bottleneck: Improving Imitation Learning, Inverse RL, and GANs by Constraining Information Flow	[slide]	-
-	ICLR2017 Deep Variational Information Bottleneck	-	-
jiangtongli	COLING2018 Modeling Multi-turn Conversation with Deep Utterance Aggregation	[slide	-

2018/12/27

Speakers	Papers	Slides	Others
yxsu	SIGHAN2018 Group Linguistic Bias Aware Neural Response Generation	[slide]	-
Shangmingyue	Arxiv2018 Dialogue Natural Language Inference	[slide]	-

2018/12/20

Speakers	Papers	Slides	Others
lixin	EMNLP2018 Semi-Supervised Learning for Neural Keyphrase Generation	[slide]	-
gaojun	ACL2018 Hierarchical Neural Story Generation	[slide]	-

2018/12/13

Speakers	Papers	Slides	Others
gaoyifan	AAAI19 A Multi-Agent Communication Framework for Question-Worthy Phrase Extraction and Question Generation	[slide]	-

2018/11/29

Speakers	Papers	Slides	Others
zhufengpan	COLING2016 Non-sentential Question Resolution using Sequence to Sequence Learning	[slide]	-
-	SIGIR2017 Incomplete Follow-up question Resolution using Retrieval based Sequence to Sequence Learning	-	[dataset]

2018/11/23

Speakers	Papers	Slides	Others
zhaoyang	ICLR2018(under review)I Know the Feeling: Learning to Converse with Empathy	[slide]	-
jcykcai	NIPS2018 Deep Generative Models with Learnable Knowledge Constraints	[slide]	-

2018/10/25

Speakers	Papers	Slides	Others
gaoyifan	ACL2018 Harvesting Paragraph-Level Question-Answer Pairs from Wikipedia	[slide]	-
shangmingyue	NIPS2017 Adversarial Ranking for Language Generation	[slide]	-
-	AAAI2018 Long Text Generation via Adversarial Training with Leaked Information	-	-

2018/10/18

Speakers	Papers	Slides	Others
gaojun	NAACL2017 Deep contextualized word representations	[slide]	-
-	Arxiv2018 Improving Language Understanding by Generative Pre-Training	-	-
-	Arxiv2018 BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding	-	-

2018/10/11

Speakers	Papers	Slides	Others
lixin	ACL2017 Neural Belief Tracker: Data-Driven Dialogue State Tracking	[slide]	-
-	ACL2018 Global-Locally Self-Attentive Encoder for Dialogue State Tracking	-	-
-	ICASSP2018 Adversarial Actor-Critic Model For Task-Completion Dialogue Policy Learning	-	-
-	ACL2018 Deep Dyna-Q: Integrating Planning for Task-Completion Dialogue Policy Learning	-	-

2018/9/21

Speakers	Papers	Slides	Others
cd	NIPS2018 Generating Informative and Diverse Conversational Responses via Adversarial Information Maximization	[slide]	-
	EMNLP2017 Sequential Matching Network-A New Architecture for Multi-turn Response Selection in Retrieval-Based Chatbots		-
zhaoyang	ACL2018 Learning to Control the Specificity in Neural Response Generation	[slide]	-

2018/8/1

Speakers	Papers	Slides	Others
gaojun	NIPS2017 Diverse and Accurate Image Description Using a Variational Auto-Encoder with an Additive Gaussian Encoding Space	[slide]	-
cd	arXiv2018 Response Generation by Context-aware Prototype Editing	[slide]	-
-	arXiv2016 Two are better than one: An ensemble of retrieval-and generation-based dialog systems	-	-
zhaoyang	AAAI2018 Dictionary-Guided Editing Networks for Paraphrase Generation	[slide]	-
ziyang	ACL2018 Learning to Ask Good Questions: Ranking Clarification Questions using Neural Expected Value of Perfect Information	[slide]	-
biwei	-	-	-
yahui	ACL2018 Token-level and sequence-level loss smoothing for RNN language models	[slide]	-
-	arXiv2018 Sounding Board: A User-Centric and Content-Driven Social Chatbot	-	-

2018/7/27 ACL Report

Speakers	Papers	Slides	Others
zhaoyang	ACL18 Report	slide	-

2018/7/18

Speakers	Papers	Slides	Others
gaojun	ACL2017 Generating Natural Answers by Incorporating Copying and Retrieving Mechanisms in Sequence-to-Sequence Learning	[slide]	-
cd	ACL18 AdvEntuRe: Adversarial Training for Textual Entailment with Knowledge-Guided Examples	[slide]	-
-	ACL18 Working Memory Networks-Augmenting Memory Networks with a Relational Reasoning Module	-	-
ziyang	IJCAI2018 SentiGAN: Generating Sentimental Texts via Mixture Adversarial Networks	[slide]	-
biwei	-	-	-
yahui	AAAI2015 Self-Paced Curriculum Learning	[slide]	-
-	ICML2018 MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels	-	-

2018/7/3 Reinforcement Learning

Speakers	Papers	Slides	Others
gaojun	AAAI2018 Flexible End-to-End Dialogue System for Knowledge Grounded Conversation	[slide]	-
cd	Nature2017 Mastering the game of Go Without human knowledge	[slide]	-
ziyang	CVPR2018 Video Captioning via Hierarchical Reinforcement Learning	[slide]	-
biwei	ICML2017 FeUdal Networks for Hierarchical Reinforcement Learning	[slide]	-
yahui	IJCAI2018 Learning to Converse with Noisy Data: Generation with Calibration	[slide]	-
-	arXiv2016 Data Distillation for Controlling Specificity in Dialogue Generation	-	-

2018/6/26 GAN review & Knowledge-incoporated Generation & RL

Speakers	Papers	Slides	Others
xiaojiang	review questions about GAN again and summarize GAN's possible use in conversation resposne geneartion.	-	-
gaojun	IJCAI2016 Neural Generative Question Answering	[slide]	-
-	AAAI2018 A Knowledge-Grounded Neural Conversation Model	-	-
cd	ACL2018 Fast Abstractive Summarization with Reinforce-Selected Sentence Rewriting	[slide]	-
-	ACL2018 Unpaired Sentiment-to-Sentiment Translation: A Cycled Reinforcement Learning Approach	-	-
yahui	NAACL2018 Discourse-Aware Neural Rewards for Coherent Text Generation	[slide]	Report of GAN
biwei	ICML2017 Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control	[slide]	-

2018/6/20 GAN

• xiaojiang's questions, hope we could have agreements on these three points, and output some reports:
  1. Why Seq2seq is better than the previous language model methods in generating language sequence. Why GAN is better than standard Seq2seq?
  2. GAN has been successfully appllied to many new image tasks, such as image generation. What are the best tasks of GAN for text?
  3. Why GAN has no break-through on text yet? All possible reasons.

Lecturers	Papers	Slides	Others
cd	Implement Adversarial Training for Text Generation (motivations and technologies)	[slide]	-
gaojun	EMNLP2017 Neural Response Generation via GAN with an Approximate Embedding Layer∗	[slide]	-
-	IJCAI2018 Commonsense Knowledge Aware Conversation Generation with Graph Attention	-	-
yahui	EMNLP2017 Adversarial Learning for Neural Dialogue Generation	[slide]	-
-	ICLR2018 MaskGAN: Better Text Generation via Filling in the __	-	-
biwei	ICML2017 Adversarial Feature Matching for Text Generation	[slide]	-