Attention over Learned Object Embeddings Enables Complex Visual Reasoning
Dec 6, 2021
Speakers
David Ding
Speaker · 0 followers
Felix Hill
Speaker · 1 follower
Adam Santoro
Speaker · 0 followers
Malcolm Reynolds
Speaker · 0 followers
Matthew Botvinick
Speaker · 0 followers
About
Neural networks have achieved success in a wide array of perceptual tasks but often fail at tasks involving both perception and higher-level reasoning. On these more challenging tasks, bespoke approaches (such as modular symbolic components, independent dynamics models or semantic parsers) targeted towards that specific type of task have typically performed better. The downside to these targeted approaches, however, is that they can be more brittle than general-purpose neural networks, requiring significant modification or even redesign according to the particular task at hand. Here, we propose a more general neural-network-based approach to dynamic visual reasoning problems that obtains state-of-the-art performance on three different domains, in each case outperforming bespoke modular approaches tailored specifically to the task. Our method relies on learned object-centric representations, self-attention and self-supervised dynamics learning, and all three elements together are required for strong performance to emerge. The success of this combination suggests that there may be no need to trade off flexibility for performance on problems involving spatio-temporal or causal-style reasoning. With the right soft biases and learning objectives in a neural network we may be able to attain the best of both worlds.Neural networks have achieved success in a wide array of perceptual tasks but often fail at tasks involving both perception and higher-level reasoning. On these more challenging tasks, bespoke approaches (such as modular symbolic components, independent dynamics models or semantic parsers) targeted towards that specific type of task have typically performed better. The downside to these targeted approaches, however, is that they can be more brittle than general-purpose neural networks, requiring…
Organizer
NeurIPS 2021
Account · 1.9k followers
Categories
AI & Data Science
Category · 10.8k presentations
About NeurIPS 2021
Neural Information Processing Systems (NeurIPS) is a multi-track machine learning and computational neuroscience conference that includes invited talks, demonstrations, symposia and oral and poster presentations of refereed papers. Following the conference, there are workshops which provide a less formal setting.
Like the format? Trust SlidesLive to capture your next event!
Professional recording and live streaming, delivered globally.
Sharing
Recommended Videos
Presentations on similar topic, category or speaker
Count-Based Temperature Scheduling for Maximum Entropy Reinforcement Learning
Watch later
Dailin Hu, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Adaptive Scheduling of Data Augmentation for Deep Reinforcement Learning
Watch later
Byungchan Ko, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Generative Active Learning for Black-Box Exploration
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
The Symbiosis of Deep Learning and Differential Equations
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Deep Reinforcement Learning
Watch later
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%
Controlled Text Generation as Continuous Optimization with Multiple Constraints
Watch later
Sachin Kumar, …
Total of 0 viewers voted for saving the presentation to eternal vault which is 0.0%