Proves that for bounded n, there exist MV embeddings with m vectors each whose Chamfer matrix requires single-vector dimension D = (ε² m)^Ω(1/ε) to approximate within ε, separating MV from SV expressiveness.
hub
On the theoretical limitations of embedding-based retrieval
24 Pith papers cite this work, alongside 3 external citations. Polarity classification is still indexing.
24
Pith papers citing it
3
external citations · external index
hub tools
citation-role summary
background 3
method 1
citation-polarity summary
representative citing papers
Proves constant-dimension 3 reachability embeddings for directed trees, O(t log n) for treewidth-t graphs, with lower bounds of Omega(n) for general DAGs and Omega(t/log(n/t)) for treewidth t, plus real-dataset results.
MaxSim similarity can exactly replicate inner products of non-negative sparse vectors of arbitrary dimension, and a proposed Signed MaxSim extension enables exact replication for arbitrary real-valued vectors.
A 0.6B LM with length-aware attention adjustments performs competitive in-context retrieval at million-token scale on MS MARCO, NQ, and LIMIT benchmarks.
TokenMinds extends Semantic ID tokenization from items to users, producing paired discrete tokens and dense embeddings via an LLM-adapted encoder-decoder for industrial recommendation.
Proves Voronoi complexity equals sign-rank for top-1 retrieval, introduces CUS diagnostic predicting retrieval failure at AUC >0.8 without labels, and AT-DW-InfoNCE objective with derived alpha^*=2.0 that improves Recall@100 on synthetic data.
Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.
A text-supervised global layout embedding augments local patch representations in late-interaction VDR, yielding +2.4 nDCG@5 and +2.3 MAP@5 gains over ColPali/ColQwen baselines on ViDoRe-v2.
Semantic Recall is a new evaluation metric for approximate nearest neighbor search that focuses only on semantically relevant results, with Tolerant Recall as a proxy when relevance labels are unavailable.
LLM-based dense retrievers generalize better when instruction-tuned but pay a specialization tax when optimized for reasoning; they resist typos and corpus poisoning better than encoder-only baselines yet remain vulnerable to semantic perturbations, with larger models and certain embedding geometry,
ARC-derived multi-agent LLM framework for safe, auditable process control with operator agents and deterministic orchestrator, evaluated on dairy ventilation.
EvoEmbedding generates evolvable embeddings via a latent memory updated during sequential processing, outperforming larger models on long-context retrieval and generalizing to 10x longer contexts in downstream tasks.
SMART unlocks latent multi-vector capabilities in single-vector embedding models by applying late interaction to frozen hidden states shaped by contrastive training, yielding consistent gains on MMEB-V2 and visual document retrieval.
Embeddings retrieve same-subfield papers at 45-52% but same-agenda papers at only 15-21%; citation rerank reaches 57-59% on agenda queries.
SIRA compresses multi-round exploratory retrieval into one corpus-discriminative BM25 action via LLM document enrichment, query-time term prediction, and corpus-statistic filtering, reporting top average performance on ten BEIR benchmarks and strong results on BrowseComp-Wikipedia without relevance
The authors introduce aspect-aware datasets GoldRiM and SilverRiM for math papers and AchGNN, a heterogeneous GNN that outperforms prior methods by jointly modeling textual semantics, citations, and author lineage across aspects.
Neural retrievers that double BM25 performance on QUEST collapse below 0.02 Recall@100 on the new LIMIT+ benchmark while lexical methods reach 0.96, with all methods degrading as compositional depth increases.
Triplet constraints realizable in D-dimensional Euclidean space cannot be preserved above 50% accuracy by any embedding of dimension at most cD for constant c<1, with UGC-hardness preventing better polynomial-time solutions in any dimension.
Generative retrieval beats dense retrieval and BM25 on the LIMIT dataset but degrades with hard negatives due to identifier ambiguity during decoding.
A mask-and-reconstruct fine-tuning objective lets autoregressive LLMs inject new knowledge without paraphrases and largely removes the reversal curse.
MetaEmbed trains fixed learnable Meta Tokens to produce granularity-organized multi-vector embeddings that support test-time scaling in multimodal retrieval.
A mean-field analysis of embedding-space crowding shows a phase transition and Fokker-Planck dynamics that drive retrieval-augmented agents to self-organize toward exclusive service of majority interests.
Argues for a denoising-first paradigm in LLM-oriented information retrieval, framing challenges via a four-stage progression and providing a taxonomy of signal-to-noise optimization techniques across the pipeline.
Foundation models stumble in pathology due to conceptual mismatches with biological tissue, requiring explicitly designed models rather than adaptations of natural-image methods.
citing papers explorer
-
LLM-Oriented Information Retrieval: A Denoising-First Perspective
Argues for a denoising-first paradigm in LLM-oriented information retrieval, framing challenges via a four-stage progression and providing a taxonomy of signal-to-noise optimization techniques across the pipeline.