Inducing Functions through Reinforcement Learning without Task Specification
This work addresses the challenge of autonomous function learning in AI, offering a novel approach inspired by biological cognition, though it appears incremental in its application to specific tasks.
The authors tackled the problem of inducing high-level cognitive functions in neural networks without explicit task specification, using a bio-inspired reinforcement learning framework, and demonstrated that functions like image classification and hidden variable estimation can emerge simultaneously without pre-training.
We report a bio-inspired framework for training a neural network through reinforcement learning to induce high level functions within the network. Based on the interpretation that animals have gained their cognitive functions such as object recognition - without ever being specifically trained for - as a result of maximizing their fitness to the environment, we place our agent in an environment where developing certain functions may facilitate decision making. The experimental results show that high level functions, such as image classification and hidden variable estimation, can be naturally and simultaneously induced without any pre-training or specifying them.