Reasoning-Modulated Representations

Neural networks leverage robust internal representations in order to generalise. Learning them is difficult, and often requires a large training set that covers the data distribution densely. We study a common setting where our task is not purely opaque. Indeed, very often we may have access to information about the underlying system (e.g. that observations must obey certain laws of physics) that any “tabula rasa” neural network would need to re-learn from scratch, penalising data efficiency. We incorporate this information into a pre-trained reasoning module, and investigate its role in shaping the discovered representations in diverse self-supervised learning settings from pixels. Our approach paves the way for a new class of data-efficient representation learning.

[1]  Jian Tang,et al.  XLVIN: eXecuted Latent Value Iteration Nets , 2020, ArXiv.

[2]  Alexander Lerchner,et al.  Spatial Broadcast Decoder: A Simple Architecture for Learning Disentangled Representations in VAEs , 2019, ArXiv.

[3]  Razvan Pascanu,et al.  Visual Interaction Networks: Learning a Physics Simulator from Video , 2017, NIPS.

[4]  Jessica B. Hamrick,et al.  Relational inductive bias for physical construction in humans and machines , 2018, CogSci.

[5]  Elise van der Pol,et al.  Contrastive Learning of Structured World Models , 2020, ICLR.

[6]  Georg Heigold,et al.  Object-Centric Learning with Slot Attention , 2020, NeurIPS.

[7]  Demis Hassabis,et al.  Improved protein structure prediction using potentials from deep learning , 2020, Nature.

[8]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[9]  Yuval Tassa,et al.  MuJoCo: A physics engine for model-based control , 2012, 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[10]  Jason Weston,et al.  Translating Embeddings for Modeling Multi-relational Data , 2013, NIPS.

[11]  Daniel Guo,et al.  Agent57: Outperforming the Atari Human Benchmark , 2020, ICML.

[12]  Mikhail Belkin,et al.  Reconciling modern machine-learning practice and the classical bias–variance trade-off , 2018, Proceedings of the National Academy of Sciences.

[13]  Demis Hassabis,et al.  Mastering Atari, Go, chess and shogi by planning with a learned model , 2019, Nature.

[14]  Charles Blundell,et al.  Neural algorithmic reasoning , 2021, Patterns.

[15]  Razvan Pascanu,et al.  Interaction Networks for Learning about Objects, Relations and Physics , 2016, NIPS.

[16]  Mark Chen,et al.  Language Models are Few-Shot Learners , 2020, NeurIPS.

[17]  Razvan Pascanu,et al.  A simple neural network module for relational reasoning , 2017, NIPS.

[18]  Marc G. Bellemare,et al.  The Arcade Learning Environment: An Evaluation Platform for General Agents , 2012, J. Artif. Intell. Res..

[19]  Alexey Dosovitskiy,et al.  Learning Object-Centric Video Models by Contrasting Sets , 2020, ArXiv.

[20]  Jürgen Schmidhuber,et al.  Relational Neural Expectation Maximization: Unsupervised Discovery of Objects and their Interactions , 2018, ICLR.

[21]  Yoshua Bengio,et al.  Unsupervised State Representation Learning in Atari , 2019, NeurIPS.

[22]  Jure Leskovec,et al.  Learning to Simulate Complex Physics with Graph Networks , 2020, ICML.

[23]  Yann LeCun,et al.  The Power and Limits of Deep Learning , 2018, Research-Technology Management.

[24]  Razvan Pascanu,et al.  Pointer Graph Networks , 2020, NeurIPS.

[25]  Samuel S. Schoenholz,et al.  Neural Message Passing for Quantum Chemistry , 2017, ICML.