TSR Task
TSR retrieval evaluates one query against a pool where multiple relevant targets may exist.
TimeSeries Retrieval Benchmark
TSRBench: Benchmarking Time Series Retrieval
TSRBench combines purpose built datasets, a modular distance library pipeline, and AdaBase scorecards to support direct comparison across leaderboard ranking, latency-accuracy trade-offs and robustness analysis.
Six figures that summarize TSRBench before deep dive
TSRBench Pipeline
TSRBench integrates data construction, distance library methods, and metric evaluation into one reproducible flow.
Datasets and Domain Shift
UCR-R and CU-RCA expose open to industrial shift with different sequence length, noise level, and incident context.
AdaBase
AdaBase mitigates single-hit dominance and captures list-wise ranking quality under pool difficulty.
Method Family Coverage
Lp distance, Elastic alignment, Correlation, Symbolic, Self supervised, and Foundation model methods are benchmarked under one protocol.
Result Preview
Results combine leaderboard, latency accuracy scatter, and AdaBase scorecard to compare ranking quality, efficiency trade-offs, and robustness in one view.
Dataset
Understand the reconstruction pipeline, filtering rules, and key dataset facts behind TSRBench.
Showcase
Browse published benchmark results and method comparisons.
Use TSRBench
Configure dataset, methods, and metrics for a focused result view.
Contact
Send concise feedback or issue reports for TSRBench.