What should a neuron aim for? Designing local objective functions based on information theory
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917027735666688 |
|---|---|
| author | Schneider, Andreas C. Neuhaus, Valentin Ehrlich, David A. Makkeh, Abdullah Ecker, Alexander S. Priesemann, Viola Wibral, Michael |
| author_facet | Schneider, Andreas C. Neuhaus, Valentin Ehrlich, David A. Makkeh, Abdullah Ecker, Alexander S. Priesemann, Viola Wibral, Michael |
| contents | In modern deep neural networks, the learning dynamics of the individual neurons is often obscure, as the networks are trained via global optimization. Conversely, biological systems build on self-organized, local learning, achieving robustness and efficiency with limited global information. We here show how self-organization between individual artificial neurons can be achieved by designing abstract bio-inspired local learning goals. These goals are parameterized using a recent extension of information theory, Partial Information Decomposition (PID), which decomposes the information that a set of information sources holds about an outcome into unique, redundant and synergistic contributions. Our framework enables neurons to locally shape the integration of information from various input classes, i.e. feedforward, feedback, and lateral, by selecting which of the three inputs should contribute uniquely, redundantly or synergistically to the output. This selection is expressed as a weighted sum of PID terms, which, for a given problem, can be directly derived from intuitive reasoning or via numerical optimization, offering a window into understanding task-relevant local information processing. Achieving neuron-level interpretability while enabling strong performance using local learning, our work advances a principled information-theoretic foundation for local learning strategies. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_02482 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | What should a neuron aim for? Designing local objective functions based on information theory Schneider, Andreas C. Neuhaus, Valentin Ehrlich, David A. Makkeh, Abdullah Ecker, Alexander S. Priesemann, Viola Wibral, Michael Information Theory Machine Learning Neural and Evolutionary Computing In modern deep neural networks, the learning dynamics of the individual neurons is often obscure, as the networks are trained via global optimization. Conversely, biological systems build on self-organized, local learning, achieving robustness and efficiency with limited global information. We here show how self-organization between individual artificial neurons can be achieved by designing abstract bio-inspired local learning goals. These goals are parameterized using a recent extension of information theory, Partial Information Decomposition (PID), which decomposes the information that a set of information sources holds about an outcome into unique, redundant and synergistic contributions. Our framework enables neurons to locally shape the integration of information from various input classes, i.e. feedforward, feedback, and lateral, by selecting which of the three inputs should contribute uniquely, redundantly or synergistically to the output. This selection is expressed as a weighted sum of PID terms, which, for a given problem, can be directly derived from intuitive reasoning or via numerical optimization, offering a window into understanding task-relevant local information processing. Achieving neuron-level interpretability while enabling strong performance using local learning, our work advances a principled information-theoretic foundation for local learning strategies. |
| title | What should a neuron aim for? Designing local objective functions based on information theory |
| topic | Information Theory Machine Learning Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2412.02482 |