Neurocoder: General-Purpose Computation Using Stored Neural Programs

Hung Le, Svetha Venkatesh
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:12204-12221, 2022.

Abstract

Artificial Neural Networks are functionally equivalent to special-purpose computers. Their inter-neuronal connection weights represent the learnt Neural Program that instructs the networks on how to compute the data. However, without storing Neural Programs, they are restricted to only one, overwriting learnt programs when trained on new data. Here we design Neurocoder, a new class of general-purpose neural networks in which the neural network “codes” itself in a data-responsive way by composing relevant programs from a set of shareable, modular programs stored in external memory. This time, a Neural Program is efficiently treated as data in memory. Integrating Neurocoder into current neural architectures, we demonstrate new capacity to learn modular programs, reuse simple programs to build complex ones, handle pattern shifts and remember old programs as new ones are learnt, and show substantial performance improvement in solving object recognition, playing video games and continual learning tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-le22b, title = {Neurocoder: General-Purpose Computation Using Stored Neural Programs}, author = {Le, Hung and Venkatesh, Svetha}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {12204--12221}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/le22b/le22b.pdf}, url = {https://proceedings.mlr.press/v162/le22b.html}, abstract = {Artificial Neural Networks are functionally equivalent to special-purpose computers. Their inter-neuronal connection weights represent the learnt Neural Program that instructs the networks on how to compute the data. However, without storing Neural Programs, they are restricted to only one, overwriting learnt programs when trained on new data. Here we design Neurocoder, a new class of general-purpose neural networks in which the neural network “codes” itself in a data-responsive way by composing relevant programs from a set of shareable, modular programs stored in external memory. This time, a Neural Program is efficiently treated as data in memory. Integrating Neurocoder into current neural architectures, we demonstrate new capacity to learn modular programs, reuse simple programs to build complex ones, handle pattern shifts and remember old programs as new ones are learnt, and show substantial performance improvement in solving object recognition, playing video games and continual learning tasks.} }
Endnote
%0 Conference Paper %T Neurocoder: General-Purpose Computation Using Stored Neural Programs %A Hung Le %A Svetha Venkatesh %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-le22b %I PMLR %P 12204--12221 %U https://proceedings.mlr.press/v162/le22b.html %V 162 %X Artificial Neural Networks are functionally equivalent to special-purpose computers. Their inter-neuronal connection weights represent the learnt Neural Program that instructs the networks on how to compute the data. However, without storing Neural Programs, they are restricted to only one, overwriting learnt programs when trained on new data. Here we design Neurocoder, a new class of general-purpose neural networks in which the neural network “codes” itself in a data-responsive way by composing relevant programs from a set of shareable, modular programs stored in external memory. This time, a Neural Program is efficiently treated as data in memory. Integrating Neurocoder into current neural architectures, we demonstrate new capacity to learn modular programs, reuse simple programs to build complex ones, handle pattern shifts and remember old programs as new ones are learnt, and show substantial performance improvement in solving object recognition, playing video games and continual learning tasks.
APA
Le, H. & Venkatesh, S.. (2022). Neurocoder: General-Purpose Computation Using Stored Neural Programs. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:12204-12221 Available from https://proceedings.mlr.press/v162/le22b.html.

Related Material