Hierarchical Data Selection via Manifold Coverage and Sparse Feature Coverage in LLM Post-training

• URL: http://arxiv.org/abs/2608.16927v1
• Date: Wed, 05 Aug 2026 04:40:26 GMT
• Title: Hierarchical Data Selection via Manifold Coverage and Sparse Feature Coverage in LLM Post-training
• Abstract summary: MASS learns low-dimensional principal manifold coordinates with a dense autoencoder for coarse semantic grouping.<n>Experiments on Vision Flan and LLaVA-CoT show that MASS consistently outperforms strong data selection baselines across multiple budgets.
• Score: 36.73565681027731
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As supervised fine-tuning data continues to scale, selecting high-value subsets from large candidate pools is crucial for reducing training cost and improving model performance. Existing methods often measure diversity directly in the original embedding space, where geometric metrics entangle dominant semantic directions, fine-grained supervision differences, and local noise. We address this limitation by formulating data selection as a coarse-to-fine hierarchical coverage problem and propose MASS. MASS learns low-dimensional principal manifold coordinates with a dense autoencoder for coarse semantic grouping, and then performs quality-aware sparse feature coverage within each group using a TopK sparse autoencoder. Experiments on Vision Flan and LLaVA-CoT show that MASS consistently outperforms strong data selection baselines across multiple budgets, and in several settings matches or surpasses full data training with only a small subset of data.

Related papers

• DecoupleMix: Decoupled Ratio Search and Convex Allocation for Scalable VLM Data Recipes [10.902458329410562]
  We formulate data construction as a systematic mixture-optimization problem.<n>For inter-class allocation, we apply a multidimensional, dataset-level assessment scoring Quality and Difficulty.<n>For intra-class composition, we apply a multidimensional, dataset-level assessment scoring Quality and Difficulty.
  arXiv  Detail & Related papers  (2026-07-27T14:55:04Z)
• CNMBI: Determining the Number of Clusters Using Center Pairwise Matching and Boundary Filtering [10.865555707958057]
  We propose an approach named CNMBI that exploits the distribution information inherent in the data space.<n>We find that different samples have different confidence levels and thereby actively remove low-confidence ones.<n> CNMBI is robust and allows for more flexibility in the dimension and shape of the target data.
  arXiv  Detail & Related papers  (2026-03-23T07:43:03Z)
• Robust Categorical Data Clustering Guided by Multi-Granular Competitive Learning [47.32771052588132]
  The nested granular cluster effect is prevalent in the implicit discrete distance space of categorical data.<n>We propose a Multi-Granular Competitiveization Learning algorithm to allow potential clusters to interactively tune themselves.<n>It is shown that the proposed MGCPL-guided Categorical Data Clustering approach is competent in exploring the nested distribution of multi-granular clusters.
  arXiv  Detail & Related papers  (2026-01-23T06:33:08Z)
• Learning from the Best, Differently: A Diversity-Driven Rethinking on Data Selection [45.327105807111934]
  Existing approaches typically rely on single or multiple-dimensional score-based selection.<n>We propose the Orthogonal Diversity-Aware Selection (ODiS) algorithm, which preserves both quality and diversity during data selection.
  arXiv  Detail & Related papers  (2025-10-21T03:37:31Z)
• Private Training & Data Generation by Clustering Embeddings [74.00687214400021]
  Differential privacy (DP) provides a robust framework for protecting individual data.<n>We introduce a novel principled method for DP synthetic image embedding generation.<n> Empirically, a simple two-layer neural network trained on synthetically generated embeddings achieves state-of-the-art (SOTA) classification accuracy.
  arXiv  Detail & Related papers  (2025-06-20T00:17:14Z)
• Adaptive and Robust DBSCAN with Multi-agent Reinforcement Learning [53.527506374566485]
  We propose a novel Adaptive and Robust DBSCAN with Multi-agent Reinforcement Learning cluster framework, namely AR-DBSCAN.<n>We show that AR-DBSCAN not only improves clustering accuracy by up to 144.1% and 175.3% in the NMI and ARI metrics, respectively, but also is capable of robustly finding dominant parameters.
  arXiv  Detail & Related papers  (2025-05-07T11:37:23Z)
• Label-independent hyperparameter-free self-supervised single-view deep subspace clustering [0.0]
  Deep subspace clustering (DSC) algorithms face several challenges that hinder their widespread adoption across domains.<n>We introduce a novel single-view DSC approach that minimizes a layer-wise self expression loss using a joint representation matrix.<n>We evaluate the proposed method on six datasets representing faces, digits, and objects.
  arXiv  Detail & Related papers  (2025-04-25T08:54:34Z)
• Unbiased Max-Min Embedding Classification for Transductive Few-Shot Learning: Clustering and Classification Are All You Need [83.10178754323955]
  Few-shot learning enables models to generalize from only a few labeled examples.<n>We propose the Unbiased Max-Min Embedding Classification (UMMEC) Method, which addresses the key challenges in few-shot learning.<n>Our method significantly improves classification performance with minimal labeled data, advancing the state-of-the-art in annotatedL.
  arXiv  Detail & Related papers  (2025-03-28T07:23:07Z)
• Adapt-$\infty$: Scalable Continual Multimodal Instruction Tuning via Dynamic Data Selection [89.42023974249122]
  Adapt-$infty$ is a new multi-way and adaptive data selection approach for lifelong instruction tuning.<n>We construct pseudo-skill clusters by grouping gradient-based sample vectors.<n>We select the best-performing data selector for each skill cluster from a pool of selector experts.<n>This data selector samples a subset of the most important samples from each skill cluster for training.
  arXiv  Detail & Related papers  (2024-10-14T15:48:09Z)
• A Framework for Fine-Tuning LLMs using Heterogeneous Feedback [69.51729152929413]
  We present a framework for fine-tuning large language models (LLMs) using heterogeneous feedback. First, we combine the heterogeneous feedback data into a single supervision format, compatible with methods like SFT and RLHF. Next, given this unified feedback dataset, we extract a high-quality and diverse subset to obtain performance increases.
  arXiv  Detail & Related papers  (2024-08-05T23:20:32Z)
• OMH: Structured Sparsity via Optimally Matched Hierarchy for Unsupervised Semantic Segmentation [69.37484603556307]
  Un Semantic segmenting (USS) involves segmenting images without relying on predefined labels. We introduce a novel approach called Optimally Matched Hierarchy (OMH) to simultaneously address the above issues. Our OMH yields better unsupervised segmentation performance compared to existing USS methods.
  arXiv  Detail & Related papers  (2024-03-11T09:46:41Z)
• Minimally Supervised Learning using Topological Projections in Self-Organizing Maps [55.31182147885694]
  We introduce a semi-supervised learning approach based on topological projections in self-organizing maps (SOMs) Our proposed method first trains SOMs on unlabeled data and then a minimal number of available labeled data points are assigned to key best matching units (BMU) Our results indicate that the proposed minimally supervised model significantly outperforms traditional regression techniques.
  arXiv  Detail & Related papers  (2024-01-12T22:51:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.