NILLI Novel Ideas in Learning-to-Learn through Interaction

All Editions

NILLI Workshop 2022 (Hybrid Mode)

Interactive environments have played a pivotal role in the development of reasoning mechanisms in intelligent species. Recent advances in language generation, multimodal learning, interactive and embodied learning with using language as a mode of instruction for learning have increased focus to addressing challenges in this growing topic of research. In the horizon, there is scope for realizing scenarios where agents with primitive task knowledge and an interact-and-learn procedure to systematically acquire knowledge through its interactions with the environment. NILLI focuses on collecting discussions to improve clarity towards the challenges in this topic of research which require modeling sophisticated continual interactive agents across diverse tasks and mediums of interactions. This interdisciplinary research topic unifies research paradigms of lifelong learning, natural language processing, embodied learning, reinforcement learning, robot learning and multi-modal learning towards building interactive and interpretable AI.

Speakers


Maja Rita Rudolph Bosch Center for AI	Karthik Narasimhan Assistant Professor, Princeton University.	Joyce Chai Professor, University of Michigan

Jakob N. Foerster Professor, Oxford University	Jacob Andreas Assistant Professor, Massachusetts Institute of Technology	Mark O. Reidl Professor,Georgia Institute of Technology

Call For Papers

We call for novel, published, unpublished or in-review articles in the format of extended abstract, short or long papers that are 2, 4 or 8 pages in length respectively. The topics of interest to the workshop are listed in the groups below:

Machine Learning for Interaction
- Modeling multi-modal and language interactions to aid continual learning.
- Interactive training for embodied agents.
- Early and negative results on learning to solve tasks through interaction.
- Non-verbal/Verbal interactive frameworks.
Interaction by and for Language
- Novel environments for language understanding through interaction.
- Language based Reinforcement Learning.
- Learning representations in grounded language.
- Language based interaction methods in interdisciplinary research.
Community Impact of Interactive Agents
- Frontiers in building interactive agents (data, frameworks, open problems).
- Applications of interactive learning in interdisciplinary research.
- Security and ethical challenges in interaction based learning.

Schedule

The workshop will be hybrid. Although the events happen online, at the conference venue there will be volunteers who can guide you to workshop viewing sites.

The underline link to find the zoom link is right here!

Times are in Abu Dhabi Time.

Opening Remarks 09:00-09:05

Invited Talk 1 ( Dr. Maja Rita Rudolph) 09:05-09:50

Lightning Talks 09:50-10:30

Revisiting the Roles of "Text” in Text Games (Yi Gu) 09:50-10:00
ComFact: A Benchmark for Linking Contextual Commonsense Knowledge (Silin Gao) 10:00-10:10
Learn What Is Possible, Then Choose What Is Best: Disentangling One-To-Many Relations in Language Through Text-based Games (Benjamin Towle) 10:10-10:20
LEMON: Language-Based Environment Manipulation via Execution-Guided Pre-training (Qi Shi) 10:20-10:30

Break 10:30-10:45

Lightning Talks 10:45-11:15

Reason first, then respond: Modular Generation for Knowledge-infused Dialogue (Leonard Adolphs) 10:45-10:55
Lexi: Self-Supervised Learning of the UI Language (Pratyay Banerjee) 10:55-11:05
Context-aware Information-theoretic Causal De-biasing for Interactive Sequence Labeling (Junda Wu) 11:05-11:15

Invited Talk 2 (Prof. Joyce Chai) 11:15-12:00

Break 12:00 - 13:00

Invited Talk 3 (Prof. Karthik Narasimhan) 13:00-13:45

Invited Talk 4 (Prof. Jakob Foerster ) 13:45-14:30

Lightning Talks 14:30-15:40

Detect-Localize-Repair: A Unified Framework for Learning to Debug with CodeT5 (Nghi D.Q. Bui) 14:30-14:40
StuBot: Learning by Teaching a Conversational Agent Through Machine Reading Comprehension (Nayoung Jin) 14:40-14:50
AERBIF: Actionable Entities Recognition Benchmark for Interactive Fiction (Ivan Yamshchikov) 14:50-15:00
Multimodal Contextualized Plan Prediction for Embodied Task Completion (Mert Inan) 15:00-15:10
Thompson sampling for interactive Bayesian optimization of dynamic masking-based language model pre-training (Iñigo Urteaga) 15:10-15:20
ReAct: Synergizing Reasoning and Acting in Language Models (Shunyu Yao) 15:20-15:30
Joint Audio/Text Training for Transformer Rescorer of Streaming Speech Recognition (Suyoun Kim) 15:30-15:40

Break 15:40-16:00

Invited Talk 5 (Prof. Jacob Andreas) 16:00-16:45

Invited Talk 6 (Prof. Mark O. Reidl ) 16:45-17:30

Closing Remarks 17:30-17:40

Accepted Papers

Direct Submissions

AERBIF: Actionable Entities Recognition Benchmark for Interactive Fiction, PDF
Ivan P. Yamshchikov, Alexey Tikhonov
Multimodal Contextualized Plan Prediction for Embodied Task Completion,PDF
Mert Inan,Aishwarya Padmakumar, Spandana Gella, Patrick Lange, Dilek Hakkani-Tur
Thompson sampling for interactive Bayesian optimization of dynamic masking-based language model pre-training,PDF
Iñigo Urteaga, Moulay Zaidane Draidia, Tomer Lancewicki, Shahram Khadivi

EMNLP-Findings

Revisiting the Roles of "Text” in Text Games,
Yi Gu, Shunyu Yao, Chuang Gan, Josh Tenenbaum and Mo Yu
ComFact: A Benchmark for Linking Contextual Commonsense Knowledge,
Silin Gao, Jena D. Hwang, Saya Kanno, Hiromi Wakaki, Yuki Mitsufuji and Antoine Bosselut

LEMON: Language-Based Environment Manipulation via Execution-Guided Pre-training,
Qi Shi, Qian Liu, Bei Chen, Yu Zhang, Ting Liu and Jian-Guang LOU

Learn What Is Possible, Then Choose What Is Best: Disentangling One-To-Many Relations in Language Through Text-based Games,
Benjamin Towle and Ke Zhou

Reason first, then respond: Modular Generation for Knowledge-infused Dialogue,
Leonard Adolphs, Kurt Shuster, Jack Urbanek, Arthur Szlam and Jason Weston

Lexi: Self-Supervised Learning of the UI Language,
Pratyay Banerjee, Shweti Mahajan, Kushal Arora, Chitta Baral and Oriana Riva

Context-aware Information-theoretic Causal De-biasing for Interactive Sequence Labeling,
Junda Wu, Rui Wang, Tong Yu, Ruiyi Zhang, Handong Zhao, Shuai Li, Ricardo Henao and Ani Nenkova

Detect-Localize-Repair: A Unified Framework for Learning to Debug with CodeT5,
Nghi D. Q. Bui, Yue Wang and Steven C.H. Hoi

StuBot: Learning by Teaching a Conversational Agent Through Machine Reading Comprehension,
Nayoung Jin and Hana Lee

ReAct: Synergizing Reasoning and Acting in Language Models,
Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, Yuan Cao

Joint Audio/Text Training for Transformer Rescorer of Streaming Speech Recognition,
Suyoun Kim, Ke Li, Lucas Kabela, Rongqing Huang, Jiedan Zhu, Ozlem Kalinli, Duc Le

Important Dates

Submission deadline: 08 23-September-2022
Notification of acceptance: 08-October-2022
Camera-ready papers due: 16-October-2022
Workshop: 8-December-2022

All deadlines are 11.59 pm UTC -12h (“anywhere on Earth”).

Submission Instructions

The submission link is Open at openreview. The style guidelines follows the format guidelines of EMNLP 2022. The workshop is non-archival but will list a PDF in the webpage after the camera-ready deadline for the ease of access for conference attendees.

Organizing Committee


Prasanna Parthasarathi Senior NLP Researcher, Huawei Noah's Ark Lab.	Koustuv Sinha Senior Ph.D. Candidate McGill University.	Kalesha Bullard Research Scientist, Google Deepmind.	Chinnadhurai Sankar Research Scientist, Meta AI.


Adina Williams Research Scientist, Meta AI.	Sarath Chandar Assistant Professor, Polytechnique de Montreal.	Marc-Alexandre Côté Principal Researcher, Microsoft Research.	Joelle Pineau Associate Professor, McGill University.

Program Committee

Nikita Moghe, University of Edinburgh
Eric Yuan, MSR
Yuanzhe Pang, New York University
Bishal Santra, IIT Kharagpur
Thao Nguyen, Brown University
Yu-Siang Wang, Googlge
Darshan Patil, University of Montreal (UdeM)
Saujas Vaduguru, Carnegie Mellon University