Frank Nielsen extends Bhattacharyya and Chernoff Bayes error bounds via quasi-arithmetic means

FrnkNlsn · x · 2026-09-13

• Frank Nielsen shares his arXiv paper (1401.4788) on bounding Bayes error, which is usually intractable to compute exactly.
• The paper restates Bayes risk via total variation distance on scaled distributions, then generalizes the Bhattacharyya and Chernoff upper-bound machinery using generalized weighted (quasi-arithmetic) means, yielding new statistical divergences and affinity coefficients as byproducts.
• New, experimentally tighter upper bounds are derived for the univariate Cauchy and multivariate t-distributions.

Original post →

More from Research

• Paper frames Transformers as discretization of continuous integro-differential equations — NandoDF · 2026-09-13
• Graph Convolutional Networks computed by hand in 12 steps: GCN is the transformer's little cousin — ProfTomYeh · 2026-09-13
• Math frontier will move with AI, but dirty proofs are unacceptable — njyx · 2026-09-13
• New Paper Demystifies RL Post-Training: Random Rewards Don't Boost LLM Capabilities — natashajaques · 2026-09-13
• One Layer Deeper Challenge Draws 15,000 Submissions, None Solves It As Intended — marksaroufim · 2026-09-13
• Intelligence Has a Speed Limit: control theory caps recursive self-improvement — docmilanfar · 2026-09-13

Research channel →