Researcher

Shi Fu

3 developments in the corpus

Following this puts new work involving it at the top of your briefing, with a note saying why it is there. Links are taken from the source record, never inferred.

Latest

Everything in the corpus, newest first.

SEP 9, 2026 · PREPRINT

A Sharp Barrier for Consistent Submodular Maximization: Any Improvement over $2-\sqrt{2}$ Entails Exponential Queries or Linear Recourse

arXiv

This is a theoretical computer science result on algorithmic complexity and approximation bounds, not a clinical or empirical study, and establishes mathematical limits rather than actionable evidence.

Insufficient InformationShi Fu Qixin Zhang Dacheng Tao

Reported

Supremum approximation ratio2−√2 ≈ 0.5858

Offline guarantee1−1/e

Algorithm convergenceO(ε−2) changes per insertion

SEP 3, 2026 · PREPRINT

Restricted Eigenvalues Beyond Gaussian Width: Threshold Occupancy under Heavy Tails

arXiv

This is a theoretical mathematical work answering a 2015 open problem by constructing counterexamples and proving worst-case sample complexity bounds; it raises and resolves a question about high-dimensional statistics rather than testing an empirical claim.

Hypothesis-GeneratingShi Fu Qixin Zhang Dacheng Tao

Reported

Gaussian sample complexityO(1 + log(1/δ))

Heavy-tailed failure regimen ≲ √(p/log p)

Sharp worst-case sample complexit…Θ(β⁻¹[d log(1/β) + log(1/δ)])

AUG 18, 2026 · PREPRINT

Feature Priming in Online Linear Regression: Sparse-Regret Lower Bounds and a Tight Univariate Rate

arXiv

Theoretical analysis of online regression algorithms with lower-bound constructions and limited empirical validation; raises questions about feature priming rather than settling clinical or practical questions.

Hypothesis-GeneratingShi Fu Qixin Zhang Dacheng Tao

Reported

Lower bound regretΩ(min{T,√d})