Meta-Learning Strategies through Value Maximization in Neural Networks
arXiv:2310.19919
Abstract
Biological and artificial learning agents face numerous choices about how to learn, ranging from hyperparameter selection to aspects of task distributions like curricula. Understanding how to make these meta-learning choices could offer normative accounts of cognitive control functions in biological learners and improve engineered systems. Yet optimal strategies remain challenging to compute in modern deep networks due to the complexity of optimizing through the entire learning process. Here we theoretically investigate optimal strategies in a tractable setting. We present a learning effort framework capable of efficiently optimizing control signals on a fully normative objective: discounted cumulative performance throughout learning. We obtain computational tractability by using average dynamical equations for gradient descent, available for simple neural network architectures. Our framework accommodates a range of meta-learning and automatic curriculum learning methods in a unified normative setting. We apply this framework to investigate the effect of approximations in common meta-learning algorithms; infer aspects of optimal curricula; and compute optimal neuronal resource allocation in a continual learning setting. Across settings, we find that control effort is most beneficial when applied to easier aspects of a task early in learning; followed by sustained effort on harder aspects. Overall, the learning effort framework provides a tractable theoretical test bed to study normative benefits of interventions in a variety of learning systems, as well as a formal account of optimal cognitive control strategies over learning trajectories posited by established theories in cognitive neuroscience.
References in corpus (18)
- Overcoming catastrophic forgetting in neural networks
- Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
- Multi-Task Learning with Deep Neural Networks: A Survey
- A mathematical theory of semantic development in deep neural networks
- Meta-Reinforcement Learning of Structured Exploration Strategies
- Lifelong Teacher-Student Network Learning
- Open-Ended Learning Leads to Generally Capable Agents
- Statistical Mechanics of Deep Linear Neural Networks: The Back-Propagating Kernel Renormalization
- Theoretical Convergence of Multi-Step Model-Agnostic Meta-Learning
- Self-Consistent Dynamical Field Theory of Kernel Evolution in Wide Neural Networks
- The Neural Race Reduction: Dynamics of Abstraction in Gated Networks
- Navigating the Trade-Off between Multi-Task Learning and Learning to Multitask in Deep Neural Networks
- Context Meta-Reinforcement Learning via Neuromodulation
- An Analytical Theory of Curriculum Learning in Teacher-Student Networks
- Continuous-Time Meta-Learning with Forward Mode Differentiation
- Loss Dynamics of Temporal Difference Reinforcement Learning
- Globally Gated Deep Linear Networks
- Memory-Based Meta-Learning on Non-Stationary Distributions