Unverified50% confidenceFactExact time
IAFTRL's regret upper bound in adversarial environments is ~O(√T), of the same order as the optimal bound for purely adversarial linear bandit algorithms.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn stochastic linear bandits, the instance-dependent optimal regret lower bound can usually be expressed in the form Ω(C* log T), matching IAFTRL's upper bound (up to logarithmic factors).75% similarUnverifiedThe follow-on trace is a core auxiliary variable in the ETD algorithm, with the recurrence formula F_{t+1} = γρ_t F_t + 1.62% similarUnverified@ai-sdk/azure reuses a large amount of the underlying logic from @ai-sdk/openai, with differentiated handling only at the levels of authentication method, endpoint configuration, and deployment name.54% similarUnverifiedGUT provides the assert_almost_eq assertion method, designed specifically for floating-point comparison, with a specifiable tolerance.54% similarUnverifiedBM25 (Best Matching 25) is a probabilistic retrieval function proposed in the Okapi system and remains the default similarity algorithm in Lucene/Elasticsearch to this day.53% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1128909API
curl https://kongchang.com/api/v1/knowledge/claims/1128909MCP
get_claim(id=1128909)