On the Local Minima of the Empirical Risk

Chi Jin, Lydia T. Liu, Rong Ge, Michael I. Jordan

Nov-20-2025, 20:31:44 GMT–Neural Information Processing Systems

Even for applications with nonconvex nonsmooth losses (such as modern deep networks), the population risk is generally significantly more well-behaved from an optimization point of view than the empirical risk.

algorithm, artificial intelligence, machine learning, (15 more...)

Neural Information Processing Systems

Nov-20-2025, 20:31:44 GMT

Conferences PDF

Country:
- North America
  - United States > California
    - Alameda County > Berkeley (0.04)
  - Canada > Quebec
    - Montreal (0.04)
- Europe
  - United Kingdom > England
    - Oxfordshire > Oxford (0.04)
  - Germany > Baden-Württemberg
    - Stuttgart Region > Stuttgart (0.04)
- Asia
  - Middle East > Jordan (0.05)
  - Afghanistan > Parwan Province
    - Charikar (0.04)

Technology:
- Information Technology > Artificial Intelligence
  - Representation & Reasoning (1.00)
  - Machine Learning > Statistical Learning
    - Gradient Descent (0.35)

Duplicate Docs Excel Report

Title
On the Local Minima of the Empirical Risk
On the Local Minima of the Empirical Risk

Similar Docs Excel Report more

Title	Similarity	Source
None found