Human memory does not preserve every moment equally. Some experiences stay vivid because they are emotionally charged, while others fade quickly.

What I am modeling

This project uses convolutional autoencoders and contrastive learning to study how physiological signals correlate with memory importance.

The goal is not just to predict whether something will be recalled. I want to understand the latent structure that turns raw experience into durable memory.

Why it matters

The same framing can be useful for continual learning systems. If we can learn when a system should keep, compress, or discard information, we get closer to memory-aware learning.

The project also gives me a concrete way to connect cognitive science ideas with neural network behavior.