Neural collapse with unconstrained features
Autor: | Dustin G. Mixon, Hans Parshall, Jianzong Pi |
---|---|
Jazyk: | angličtina |
Rok vydání: | 2020 |
Předmět: |
FOS: Computer and information sciences
Computational Mathematics Computer Science - Machine Learning Algebra and Number Theory Quantitative Biology::Neurons and Cognition Signal Processing Computer Science::Neural and Evolutionary Computation Radiology Nuclear Medicine and imaging Analysis Machine Learning (cs.LG) |
Popis: | Neural collapse is an emergent phenomenon in deep learning that was recently discovered by Papyan, Han and Donoho. We propose a simple "unconstrained features model" in which neural collapse also emerges empirically. By studying this model, we provide some explanation for the emergence of neural collapse in terms of the landscape of empirical risk. |
Databáze: | OpenAIRE |
Externí odkaz: |