Recurrent Neural Networks and Related Machines That Learn Algorithms

The Neural Data Router: Adaptive Control Flow in Transformers Improves Systematic Generalization

Auteurs: R Csordás, K Irie, J Schmidhuber
Publié dans: ICLR 2022, 2022
Éditeur: ICLR 2022

Linear Transformers Are Secretly Fast Weight Programmers

Auteurs: I. Schlag*, K. Irie*, J. Schmidhuber
Publié dans: ICML 2021, Proceedings of the 38th International Conference on Machine Learning, 2021
Éditeur: ICML 2021

Parameter-based value functions

Auteurs: F. Faccio, L. Kirsch, J. Schmidhuber
Publié dans: NeurIPS 2020 Workshop on Offline Reinforcement Learning, 2020
Éditeur: NeurIPS 2020 Workshop on Offline Reinforcement Learning

Going Beyond Linear Transformers with Recurrent Fast Weight Programmers

Auteurs: K. Irie*, I. Schlag*, R. Csordás, J. Schmidhuber
Publié dans: NeurIPS 2021, 2021
Éditeur: NeurIPS 2021

Training and Generating Neural Networks in Compressed Weight Space

Auteurs: K. Irie, J. Schmidhuber
Publié dans: ICLR 2021 Workshop on Neural Compression, 2021
Éditeur: ICLR 2021 Workshop on Neural Compression

A Modern Self-Referential Weight Matrix That Learns to Modify Itself

Auteurs: K. Irie, I. Schlag, R. Csordás, J. Schmidhuber
Publié dans: NeurIPS 2021 Workshop on Deep Reinforcement Learning, 2021
Éditeur: NeurIPS 2021 Workshop on Deep Reinforcement Learning

Reward-Weighted Regression Converges to a Global Optimum

Auteurs: M. Štrupl*, F. Faccio*, D. R. Ashley, R. K. Srivastava, J. Schmidhuber
Publié dans: AAAI 2022, in press, 2022
Éditeur: AAAI 2022

Learning Associative Inference Using Fast Weight Memory

Auteurs: I. Schlag, T. Munkhdalai, J. Schmidhuber
Publié dans: ICLR 2021, 2021
Éditeur: ICLR 2021

The Devil is in the Detail: Simple Tricks Improve Systematic Generalization of Transformers

Auteurs: R Csordás, K Irie, J Schmidhuber
Publié dans: EMNLP 2021, Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, 2021, Page(s) 619–634
Éditeur: EMNLP 2021

Bayesian Brains and the Rényi Divergence

Auteurs: N. Sajid*, F. Faccio*, L. Da Costa, T. Parr, J. Schmidhuber, K. Friston
Publié dans: CNS*2021, 2021
Éditeur: CNS*2021

Improving Stateful Premise Selection with Transformers

Auteurs: K. Prorokovic, M. Wand, J. Schmidhuber
Publié dans: CICM 2021, 2021
Éditeur: CICM 2021

Improving Baselines in the Wild

Auteurs: K. Irie, I. Schlag, R. Csordás, J. Schmidhuber
Publié dans: NeurIPS 2021 Workshop on Distribution Shifts, 2021
Éditeur: NeurIPS 2021 Workshop on Distribution Shifts

Unsupervised Object Keypoint Learning using Local Spatial Predictability

Auteurs: A. Gopalakrishnan, S. van Steenkiste, J. Schmidhuber
Publié dans: ICML 2020 workshop on Object-Oriented Learning: Perception, Representation and Reasoning, 2020
Éditeur: ICML

Parameter-based value functions

Auteurs: F. Faccio, L. Kirsch, J. Schmidhuber
Publié dans: ICLR 2021, 2021
Éditeur: ICLR 2021

Reward-Weighted Regression Converges to a Global Optimum

Auteurs: M. Štrupl*, F. Faccio*, D. R. Ashley, R. K. Srivastava, J. Schmidhuber
Publié dans: ICML 2021 Workshop on Reinforcement Learning Theory, 2021
Éditeur: ICML 2021 Workshop on Reinforcement Learning Theory

Policy Optimization via Importance Sampling

Auteurs: A. M. Metelli, M. Papini, F. Faccio, M. Restelli
Publié dans: NeurIPS 2018, 2018
Éditeur: published at the Neural Information Processing Systems Conference

Improving Differentiable Neural Computers Through Memory Masking, De-allocation, and Link Distribution Sharpness Control

Auteurs: R. Csordas, J. Schmidhuber
Publié dans: ICLR 2019, 2019
Éditeur: Published as a conference paper at ICLR 2019

Modular Networks: Learning to Decompose Neural Computation

Auteurs: L. Kirsch, J. Kunze, D. Barber
Publié dans: NeurIPS 2018, 2018
Éditeur: published at the Neural Information Processing Systems Conference

Learning to Reason with Third Order Tensor Products

Auteurs: I. Schlag, J. Schmidhuber
Publié dans: NeurIPS 2018, 2018
Éditeur: published at the Neural Information Processing Systems Conference

Recurrent World Models Facilitate Policy Evolution

Auteurs: D. Ha, J. Schmidhuber
Publié dans: NeurIPS 2018, 2018
Éditeur: published at the Neural Information Processing Systems Conference

Enhancing the Transformer with Explicit Relational Encoding for Math Problem Solving

Auteurs: I. Schlag, P. Smolensky, R. Fernandez, N. Jojic, J. Schmidhuber, J. Gao
Publié dans: Under review, 2020
Éditeur: Not disclosed yet

Improving Generalization in Meta Reinforcement Learning using Neural Objectives

Auteurs: L. Kirsch, S. van Steenkiste, J. Schmidhuber
Publié dans: ICLR 2020, 2020
Éditeur: Published as a conference paper at ICLR 2020

Learning Adaptive Control Flow in Transformers for Improved Systematic Generalization

Auteurs: R. Csordás, K. Irie, J. Schmidhuber
Publié dans: NeurIPS 2021 Workshop on Advances in Programming Languages and Neurosymbolic Systems, 2021
Éditeur: NeurIPS 2021 Workshop on Advances in Programming Languages and Neurosymbolic Systems

Augmenting Classic Algorithms with Neural Components for Strong Generalisation on Ambiguous and High-Dimensional Data

Auteurs: I. Schlag, J. Schmidhuber
Publié dans: NeurIPS 2021 Workshop on Advances in Programming Languages and Neurosymbolic Systems, 2021
Éditeur: NeurIPS 2021 Workshop on Advances in Programming Languages and Neurosymbolic Systems

Parameter-based value functions

Auteurs: F. Faccio, L. Kirsch, J. Schmidhuber
Publié dans: NeurIPS 2020 Workshop on Deep Reinforcement Learning, 2021
Éditeur: NeurIPS 2020 Workshop on Deep Reinforcement Learning

Unsupervised Object Keypoint Learning using Local Spatial Predictability

Auteurs: A. Gopalakrishnan, S. van Steenkiste, J. Schmidhuber
Publié dans: ICLR 2021, 2021
Éditeur: ICLR 2021

Are Neural Nets Modular? Inspecting Functional Modularity Through Differentiable Weight Masks

Auteurs: R Csordás, S van Steenkiste, J Schmidhuber
Publié dans: ICLR 2021, 2021
Éditeur: ICLR 2021

Exploring through Random Curiosity with General Value Functions

Auteurs: A. Ramesh, L. Kirsch, S. van Steenkiste, J. Schmidhuber
Publié dans: NeurIPS 2021 Workshop on Deep Reinforcement Learning, 2021
Éditeur: NeurIPS 2021 Workshop on Deep Reinforcement Learning

Meta learning backpropagation and improving it

Auteurs: L. Kirsch, J. Schmidhuber
Publié dans: NeurIPS 2021, 2021
Éditeur: NeurIPS 2021

Bayesian Brains and the Rényi Divergence

Auteurs: N. Sajid*, F. Faccio*, L. Da Costa, T. Parr, J. Schmidhuber, K. Friston
Publié dans: Neural Computation, Numéro 34, 829–855 (2022), 2021, ISSN 1530-888X
Éditeur: Neural computation, Cambridge: MIT Press

Generative Adversarial Networks are Special Cases of Artificial Curiosity (1990) and also Closely Related to Predictability Minimization (1991)

Auteurs: J. Schmidhuber
Publié dans: Neural Networks, 2020, ISSN 0893-6080
Éditeur: Pergamon Press Ltd.

Automatic Embedding of Stories Into Collections of Independent Media

Auteurs: D. R. Ashley*, V. Herrmann*, Z. Friggstad, K. W. Mathewson, J. Schmidhuber
Publié dans: ArXiv, 2021
Éditeur: ArXiv

All You Need Is Supervised Learning: From Imitation Learning to Meta-RL With Upside Down RL

Auteurs: K. Arulkumaran, D. R. Ashley, J. Schmidhuber, R. K. Srivastava
Publié dans: RLDM 2022, 2022
Éditeur: RLDM 2022

Learning Relative Return Policies With Upside-Down Reinforcement Learning

Auteurs: D. R. Ashley, K. Arulkumaran, J. Schmidhuber, R. K. Srivastava
Publié dans: RLDM 2022, 2022
Éditeur: RLDM 2022

Publications

Partager cette page

Télécharger