The Thirty Sixth Conference on Neural Information Processing Systems (NeurIPS)

Dec 2, 2022

https://attention-learning-workshop.github.io/

When: Dec 2, 2022 9AM – 6PM (local time, UTC-06:00) 

Where: Room 399 (in-person) or (virtually) https://neurips.cc/virtual/2022/workshop/49996 (requires NeurIPS registration)

WORKSHOP DETAILS

The All Things Attention workshop aims to foster connections across disparate academic communities that conceptualize “Attention” such as Neuroscience, Psychology, Machine Learning, and Human-Computer Interaction. Workshop topics of interest include (but are not limited to):

1. Relationships between biological and artificial attention

1. What are the connections between different forms of attention in the human brain and present deep neural network architectures? 

2. Can the anatomy of human attention models provide useful insights to researchers designing architectures for artificial systems? 

3. Given the same task and learning objective, do machines learn attention mechanisms that are different from humans? 

2. Attention for reinforcement learning and decision making

1. How have reinforcement learning agents leveraged attention in decision making?

2. Do decision-making agents today have implicit or explicit formalisms of attention?

3. How can AI agents build notions of attention without explicitly baked in notions of attention?

4. Can attention significantly enable AI agents to scale e.g. through gains in sample efficiency, and generalization?

3. Benefits and formulation of attention mechanisms for continual / lifelong learning

1. How can continual learning agents optimize for retention of knowledge for tasks that it already learned? 

2. How can the amount of interference between different inputs be controlled via attention? 

3. How does the executive control of attention evolve with learning in humans? 

4. How can we study the development of attentional systems in infancy and childhood to better understand how attention can be learned?

4. Attention as a tool for interpretation and explanation

1. How have researchers leveraged attention as a visualization tool?

2. What are the common approaches when using attention as a tool for interpretability in AI? 

3. What are the major bottlenecks and common pitfalls in leveraging attention as a key tool for explaining the decisions of AI agents?

4. How can we do better?

5. The role of attention in human-computer interaction and human-robot interaction

1. How do we detect aspects of human attention during interactions, from sensing to processing to representations?   

2. What systems benefit from human attention modeling, and how do they use these models?

3. How can systems influence a user’s attention, and what systems benefit from this capability?

4. How can a system communicate or simulate its own attention (humanlike or algorithmic) in an interaction, and to what benefit?

5. How do attention models affect different applications, like collaboration or assistance, in different domains, like autonomous vehicles and driver assistance systems, learning from demonstration, joint attention in collaborative tasks, social interaction, etc.?

6. How should researchers thinking about attention in different biological and computational fields organize the collection of human gaze data sets, modeling gaze behaviors, and utilizing gaze information in various applications for knowledge transfer and cross-pollination of ideas?

6. Attention mechanisms in Deep Neural Network (DNN) architectures

1. How does attention in DNN such as transformers relate to existing formalisms of attention in cogsci/psychology? 

2. Do we have a concrete understanding of how and if self-attention in transformers contributes to its vast success in recent models such as GPT2, GPT3, DALLE.? 

3. Can our understanding of attention from other fields inform the progress we have achieved in recent breakthroughs?

CONFIRMED SPEAKERS & PANELISTS

Speakers:

Pieter Roelfsema (Netherlands Institute for Neuroscience)

James Whittington (University of Oxford)

Ida Momennejad (Microsoft Research)

Erin Grant (UC Berkeley)

Henny Admoni (Carnegie Mellon University)

Tobias Gerstenberg (Stanford University)

Vidhya Navalpakkam (Google Research)

Shalini De Mello (NVIDIA)

Panelists:

David Ha (Google Brain)

Pieter Roelfsema (Netherlands Institute for Neuroscience)

James Whittington (University of Oxford)

Ida Momennejad (Microsoft Research)

Henny Admoni (Carnegie Mellon University)

Tobias Gerstenberg (Stanford University)

Shalini De Mello (NVIDIA)

Vidhya Navalpakkam (Google Research)

Erin Grant (UC Berkeley)

Ramakrishna Vedantam (Meta AI Research)

Megan deBettencourt (University of Chicago)

Cyril Zhang (Microsoft Research)

ORGANIZERS

Akanksha Saran (Microsoft Research, NYC)

Khimya Khetarpal (McGill University, Mila Montreal)

Reuben Aronson (Carnegie Mellon University)

Abhijat Biswas (Carnegie Mellon University)

Ruohan Zhang (Stanford University)

Grace Lindsay (University College London, New York University)

Scott Neikum (University of Texas at Austin, University of Massachusetts)

CONTACT

Please reach out to us at attention-workshop@googlegroups.com  if you have any questions. We look forward to receiving your submissions!

Kind Regards,

Workshop Organizers

All things Attention- Bridging different perspectives on attention