Self-building Neural Networks

Abstract

During the first part of life, the brain develops while it learns through a process called synaptogenesis. The neurons, growing and interacting with each other, create synapses. However, eventually the brain prunes those synapses. While previous work focused on learning and pruning independently, in this work we propose a biologically plausible model that, thanks to a combination of Hebbian learning and pruning, aims to simulate the synaptogenesis process. In this way, while learning how to solve the task, the agent translates its experience into a particular network structure. Namely, the network structure builds itself during the execution of the task. We call this approach Self-building Neural Network (SBNN). We compare our proposed SBNN with traditional neural networks (NNs) over three classical control tasks from OpenAI. The results show that our model performs generally better than traditional NNs. Moreover, we observe that the performance decay while increasing the pruning rate is smaller in our model than with NNs. Finally, we perform a validation test, testing the models over tasks unseen during the learning phase. In this case, the results show that SBNNs can adapt to new tasks better than the traditional NNs, especially when over $80\%$ of the weights are pruned.

arXiv commentTo appear in the Genetic and Evolutionary Computation Conference Companion (GECCO '23 Companion) Proceedings, July 15--19, 2023, Lisbon, Portugal

References in corpus (10)

• The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
• The CMA Evolution Strategy: A Tutorial
• OpenAI Gym
• The State of Sparsity in Deep Neural Networks
• Sparsity in Deep Learning: Pruning and growth for efficient inference and training in neural networks
• CMA-ES for Hyperparameter Optimization of Deep Neural Networks
• Born to Learn: the Inspiration, Progress, and Future of Evolved Plastic Artificial Neural Networks
• Covariance Matrix Adaptation for the Rapid Illumination of Behavior Space
• A topological insight into restricted Boltzmann machines
• Hebbian Deep Learning Without Feedback