VASTRA:依查詢複雜度調控的高效代理式 Top-K 影片檢索
VASTRA: Efficient Agentic Top-K Interactive Video Analysis
研究速讀
作者提出 VASTRA,一套代理式影片檢索系統,主張以查詢複雜度而非開放式推理迴圈來決定執行流程:簡單查詢直接由預先建立的摘要排序,較難查詢才在真正不確定處進行針對性驗證,並稱在貪婪服務下可重現排序結果。摘要報告在 40 條基準查詢上,快速模式較最佳 ReAct 代理快 12.2 倍、成本低 10.4 倍,但 mAP 少 5.0 點。
值得留意
- 控制流程外置於模型之外,而非藏於不斷增長的思維鏈中,作者稱此設計可從構造上達成確定性控制流程。
- VASTRA-Fast 在 40 條基準查詢上耗時 4.2 秒、成本 0.0044 美元,相較最佳 ReAct 代理的 51.2 秒與 0.0458 美元。
- VASTRA-Max 的 mAP 為 0.653,作者稱與 ReAct 代理的 0.649 相當,但花費較高且宣稱無逐次執行變異。
閱讀限制
僅閱讀標題與摘要,未檢視方法細節、基準查詢組成、成本計算方式或同儕審查狀態;所有數據均為作者自述,未經獨立驗證。
原始摘要 [EN]
Vision-language models (VLMs) enable natural language video retrieval, bypassing traditional hand-engineered pipelines. Integrating agentic reasoning frameworks (e.g., ReAct) further advances this paradigm by allowing models to dynamically inspect video content rather than relying on a single visual pass. Applying this pattern naively to video retrieval inherits problems that video systems cannot afford: results vary from run to run under identical conditions, and cost and latency scale with every reasoning step the agent takes rather than with the work the query actually requires. We present VASTRA, an agentic video retrieval system whose execution is dictated by query complexity rather than an open-ended reasoning loop: cheap, reliably answerable queries are ranked directly from pre-ingested summaries, while harder queries trigger targeted verification only where relevance is genuinely uncertain. This keeps control flow external to the model rather than latent in a growing chain of thought, establishing deterministic control flow by construction and end-to-end rank reproducibility under greedy serving. Across 40 benchmark queries, VASTRA exposes explicit presets: VASTRA-Fast runs 12.2x faster (4.2 s vs. 51.2 s) and 10.4x cheaper ($0.0044 vs. $0.0458) than best performing ReAct agent at a 5.0-point mAP reduction, while VASTRA-Max matches peak ReAct agent accuracy (0.653 vs. 0.649 mAP) at comparable latency, trading higher spend ($0.1022 vs. $0.0458) for zero run-to-run variance.
AI 摘要僅根據公開標題與摘要整理;研究結論來自作者,請以原文為準。