Generalization—the ability to perform well beyond the training context—is a hallmark of biological and artificial intelligence, yet anticipating unseen failures remains a central challenge. Conventional approaches often take a “bottom-up” m…
Our knowledge can only be finite, while our ignorance must necessarily be infinite. - Karl Popper A place to record my feeling, my thought, my exploration, and my growth. Most of the posts here are translated with the help of ChatGPT. Click…
Our knowledge can only be finite, while our ignorance must necessarily be infinite. - Karl Popper (Translated from the Mandarin version with the help of GPT-5.1) Taking advantage of the office’s annual renovation closure at the end of Augus…
Integrating task-relevant information into neural representations is a fundamental ability of both biological and artificial intelligence systems. Recent theories have categorized learning into two regimes: the rich regime, where neural net…
Our knowledge can only be finite, while our ignorance must necessarily be infinite. - Karl Popper (Translated from the Mandarin version with the help of GPT-4o) It’s been nearly five or six years since I started mingling with physicists. I
From an eagle spotting a fish in shimmering water to a scientist extracting patterns from noisy data, many cognitive tasks require untangling overlapping signals. Neural circuits achieve this by transforming complex sensory inputs into dist…