SEA: Search Engines Amsterdam
Details
This meetup session will focus on SIGIR 2026 and feature three speakers from University of Amsterdam: Oscar Ramirez Milian, Jingwei Kang, and Yongkang Li. They will present their papers accepted at SIGIR 2026.
Location: Science Park 904, Room C3.161
Date: Friday, May 29
Time: 16:00-17:00
Zoom link: https://uva-live.zoom.us/j/65011610507
Details below:
Speaker #1: Oscar Ramirez Milian from University of Amsterdam
Title: TBD
Abstract: TBD
Bio: TBD
Speaker #2: Jingwei Kang from University of Amsterdam
Title: TBD
Abstract: TBD
Bio:TBD
Speaker #3: Yongkang Li from University of Amsterdam
Title: Spectral Tempering for Embedding Compression in Dense Passage Retrieval
Abstract: Dimensionality reduction is critical for deploying dense retrieval systems at scale, yet mainstream post-hoc methods face a fundamental trade-off: principal component analysis (PCA) preserves dominant variance but underutilizes representational capacity, while whitening enforces isotropy at the cost of amplifying noise in the heavy-tailed eigenspectrum of retrieval embeddings. Intermediate spectral scaling methods unify these extremes by reweighting dimensions with a power coefficient gamma, but treat gamma as a fixed hyperparameter that requires task-specific tuning. We show that the optimal scaling strength gamma is not a global constant: it varies systematically with target dimensionality k and is governed by the signal-to-noise ratio (SNR) of the retained subspace. Based on this insight, we propose Spectral Tempering (SpecTemp), a learning-free method that derives an adaptive gamma(k) directly from the corpus eigenspectrum using local SNR analysis and knee-point normalization, requiring no labeled data or validation-based search. Extensive experiments demonstrate that Spectral Tempering consistently achieves near-oracle performance relative to grid-searched gamma^(k) while remaining fully learning-free and model-agnostic.
Bio: Yongkang is a third-year PhD student at the IR Lab at the University of Amsterdam. He is very interested in embedding-based retrieval systems. His main research focuses on the robustness and generalization of dense retrieval.
Counter: SEA Talks #305, #306 and #307.