Inductive biases for deep learning of higher-level cognition

A Goyal, Y Bengio - Proceedings of the Royal Society A, 2022 - royalsocietypublishing.org
A fascinating hypothesis is that human and animal intelligence could be explained by a few
principles (rather than an encyclopaedic list of heuristics). If that hypothesis was correct, we …

The relational bottleneck as an inductive bias for efficient abstraction

TW Webb, SM Frankland, A Altabaa, S Segert… - Trends in Cognitive …, 2024 - cell.com
A central challenge for cognitive science is to explain how abstract concepts are acquired
from limited experience. This has often been framed in terms of a dichotomy between …

Conditional object-centric learning from video

T Kipf, GF Elsayed, A Mahendran, A Stone… - ar**
X Wen, B Zhao, A Zheng… - Advances in neural …, 2022 - proceedings.neurips.cc
In this paper, we tackle the problem of learning visual representations from unlabeled scene-
centric data. Existing works have demonstrated the potential of utilizing the underlying …

Improving coherence and consistency in neural sequence models with dual-system, neuro-symbolic reasoning

M Nye, M Tessler, J Tenenbaum… - Advances in Neural …, 2021 - proceedings.neurips.cc
Human reasoning can be understood as an interplay between two systems: the intuitive and
associative (" System 1") and the deliberative and logical (" System 2"). Neural sequence …

Weakly supervised representation learning with sparse perturbations

K Ahuja, JS Hartford, Y Bengio - Advances in Neural …, 2022 - proceedings.neurips.cc
The theory of representation learning aims to build methods that provably invert the data
generating process with minimal domain knowledge or any source of supervision. Most prior …