Yukun Yang

The paper develops a neural network model integrating cognitive maps, stochastic computing, and compositional coding to enable flexible planning and problem-solving in energy-efficient hardware.

This paper develops a lightweight algorithm inspired by brain's contextual encoding of episodic memories for efficient goal-directed action selection.

The paper introduces BTSP, a mechanism for binding distributed representations that enables flexible retrieval, and demonstrates its applications in visual encoding, hierarchical binding, and NLP.

The paper links cognitive maps, successor-relation based reinforcement learning, and BTSP by showing how neural codes encode sequential structure and enable plan generation.

This paper presents a theoretical framework showing neurons can learn and represent order relations through synaptic plasticity, explaining multiple experimental findings and enabling efficient inference.

The paper shows that local synaptic plasticity enables neural networks to learn cognitive maps for planning, achieving performance comparable to AI offline planning algorithms without requiring backpropagation or large datasets.