Shiyan Liu

Author directory

2026

Dense retrievers excel at first-stage candidate generation but lack effective reranking in zero-resource settings. Existing approaches face a fundamental dilemma: cross-encoders deliver strong reranking quality but require costly supervised training and incur high latency, while unsupervised BM25 reranking consistently degrades dense retrieval performance on most of BEIR benchmarks. We propose DART (Dense Adaptive Reranking at Test-time), which resolves this dilemma by adapting the scoring function at inference time. For each query, the top-ranked documents serve as pseudo-positive examples and the bottom-ranked as pseudo-negative examples, providing noisy but readily available supervision to adapt a bilinear scoring matrix W via a small number of gradient updates. We further introduce a confidence-weighted margin loss and a cross-query momentum buffer that warm-starts adaptation across queries. On six BEIR benchmarks, DART achieves a mean per-dataset relative NDCG@10 gain of +2.1% over the dense retrieval baseline with under 10ms additional latency per query, demonstrating a powerful capability for zero-shot performance enhancement and cross-domain generalization.
Automatic prompt optimization (APO) has emerged as a powerful paradigm for improving LLM performance without manual prompt engineering. Reflective APO methods such as GEPA iteratively refine prompts by diagnosing failure cases, but the optimization process remains black-box and label-free, leading to uninterpretable trajectories and systematic failure. We identify and empirically demonstrate four limitations: on GSM8K with a defective seed, GEPA degrades accuracy from 23.81% to 13.50%. We propose VISTA, a multi-agent APO framework that decouples hypothesis generation from prompt rewriting, enabling semantically labeled hypotheses, parallel minibatch verification, and interpretable optimization trace. A two-layer explore-exploit mechanism combining random restart and epsilon-greedy sampling further escapes local optima. VISTA recovers accuracy to 87.57% on the same defective seed and consistently outperforms baselines across all conditions on GSM8K and AIME2025.
Graduate program selection is a high-stakes decision where applicants struggle to identify programs matching their latent academic profiles. We address this as a retrieval problem: given a student profile, rank 531 graduate programs by estimated admission compatibility, i.e., how closely the student’s profile resembles each program’s historical admit pool. Because applicants rarely report rejections, training data consists only of positive labels, making negative supervision unavailable. We propose PURE (Positive-Unlabeled Retrieval Encoder), a positive-unlabeled biencoder with InfoNCE contrastive loss and a gated structural residual that fuses GPA and experience counts with free-text descriptions. With a novel multi-positive MRR protocol that corrects for false negatives, PURE achieves MRR=0.2696, outperforming BM25 (+4.9 points) and GPT-4o-mini (+8.3 points), while remaining statistically indistinguishable from Biencoder + GPT-4o-mini reranking at negligible cost.
Search