Yong-Hoon Choi, Youngjin Cho
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. If it links a repository, it is listed below.
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Historical retrieval for time-series prediction commonly treats past similarity as a proxy for usefulness. We ask a different question: which historical examples should be expected to matter for a query? We define predictive relevance as expected future utility conditioned on information observable at inference, and use realized futures only during training as privileged supervision. A normalizedpattern retriever first forms a coarse candidate set, and a lightweight residual MLP learns a listwise future-compatibility target while remaining strictly past-only at test time. Our method does not replace similarity-based candidate generation; it asks whether similarity-selected candidates can be ranked by a more predictive final relevance criterion. Optimal relevance further decomposes into candidate-level utility and query-specific compatibility, motivating Candidate-Prior and Shuffled-Future controls. Across six benchmarks, the reranker improves Pattern retrieval while revealing candidate-global, query-specific, and mixed relevance regimes. On 12 confirmatory tasks it improves Pattern in all 12 and outperforms a matchedprotocol SARAF retrieval rule in all 12. Architecture-matched ablations show that correct future supervision, rather than the MLP or added context alone, drives gains in query-specific regimes. Alternative-similarity experiments provide an important counterpoint: a strong last-value-anchored L2 rule remains superior in some domains, whereas future-supervised relevance is strongest in query-specific regimes such as Solar. Candidate-pool diagnostics show that this contrast is not explained solely by coarse Pattern retrieval. The resulting picture is not universal superiority of one retriever, but a structured, domain-dependent notion of historical relevance.
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2608.23221")
get_code_for_paper("2608.23221")
have("2608.23221")
Connect an agent — have() is free.