FeatureHospital: A Skill-Driven Multi-Agent Framework for Automated Algorithm Customization in Multi-View Multi-Label Feature Selection

Junxuan Li; Zhiqi Chen; Yuzhou Liu; Peng Zhang; Huaxiao Liu

cs.AIAug 17, 2026

FeatureHospital: A Skill-Driven Multi-Agent Framework for Automated Algorithm Customization in Multi-View Multi-Label Feature Selection

Authors: Junxuan Li, Zhiqi Chen, Yuzhou Liu, Peng Zhang, Huaxiao Liu

Organizations: College of Computer Science and Technology, Jilin University, Jilin, China

Abstract

Multi-view multi-label feature selection aims to identify a compact and informative feature subset from heterogeneous views while preserving discriminative information for multiple labels. Existing methods are generally developed from specific modeling perspectives and incorporate mechanisms tailored to particular data characteristics. Designing suitable feature selection algorithms across datasets with diverse and heterogeneous characteristics still relies heavily on expert knowledge and substantial manual effort, imposing considerable time and labor costs that severely hinder the practical adoption of feature selection. To address this problem, we propose FeatureHospital, a Skill-driven multi-agent framework for automated multi-view multi-label feature selection algorithm design. FeatureHospital first diagnoses the target dataset to identify its feature selection issues. Based on the diagnosis, specialist agents equipped with domain Skills then prescribe corresponding optimization strategies and Loss terms for different issues. After that, the resulting prescriptions are reconciled to remove overlaps and resolve conflicts before being integrated into a compact dataset-specific objective. Finally, the constructed objective is optimized to select the final feature subset. Experimental results demonstrate that FeatureHospital can construct effective feature selection algorithms for different datasets based on their individual characteristics.

Figures & tables

Figure 1: Conceptual comparison between conventional feature selection algorithm design and FeatureHospital. (A) In the conventional workflow, researchers manually design a feature selection algorithm for a particular dataset, and the same algorithm may become ineffective when applied to a new dataset, requiring repeated analysis and redesign. (B) FeatureHospital treats each dataset as a patient, diagnoses its feature selection issues, and employs Skill-equipped agents to automatically construct a dataset-specific feature selection algorithm.

Figure 2: Overview of FeatureHospital. The framework proceeds in four stages: (1) analyze the target dataset and identify potential problems affecting feature selection; (2) categorize the identified problems and assign them to suitable specialist agents; (3) select corresponding loss functions for each problem; (4) remove redundant or conflicting loss functions, balance their contributions, and combine them into a dataset-specific objective; and finally optimize the constructed objective to select features and evaluate the resulting feature subset.

Dataset	Domain	V / n / d / l	Views (dimensionality)
SCENE	Image	5/4400/634/33	CH(64), CM(225), CORR(144), EDH(73), WT(128)
yeast	Biology	2/2417/103/14	GE(79), PP(24)
VOC07	Image	3/3817/712/20	DH(100), GIST(512), HH(100)
MIRFlickr	Image	3/4053/712/38	DH(100), GIST(512), HH(100)
mfeat	Digits	6/2000/649/10	FOU(76), FAC(216), KAR(64), PIX(240), ZER(47), MOR(6)
emotions	Music	2/593/72/6	RHY(8), TIM(64)

Table 1: Detailed information of the datasets used in our experiments. The statistics column reports the number of views, instances, features, and labels, respectively.

Dataset	DHLI	EF 2 FS	ENM	GRAFS	I 2 VSLC	MSFS	LLM-Select	FeatureHospital
AP $\uparrow$
SCENE	0.7958 ± 0.0042	0.7678 ± 0.0033	0.7837 ± 0.0044	0.8058 ± 0.0043	0.8064 ± 0.0047	0.7681 ± 0.0240	0.8069 ± 0.0041	0.8011 ± 0.0041
yeast	0.6893 ± 0.0054	0.6788 ± 0.0090	0.6870 ± 0.0085	0.6895 ± 0.0036	0.6917 ± 0.0063	0.6468 ± 0.0112	0.7016 ± 0.0047	0.6976 ± 0.0081
VOC07	0.5885 ± 0.0081	0.5067 ± 0.0116	0.5941 ± 0.0070	0.5858 ± 0.0052	0.6003 ± 0.0032	0.5082 ± 0.0103	0.6070 ± 0.0049	0.6073 ± 0.0042
MIRFlickr	0.6621 ± 0.0060	0.6706 ± 0.0058	0.7025 ± 0.0048	0.6741 ± 0.0038	0.6634 ± 0.0059	0.5606 ± 0.0465	0.7027 ± 0.0039	0.7011 ± 0.0048
mfeat	0.6720 ± 0.0093	0.4397 ± 0.1311	0.8355 ± 0.0384	0.8925 ± 0.0109	0.9187 ± 0.0084	0.7461 ± 0.0978	0.8492 ± 0.0065	0.9360 ± 0.0056

Table 2: Comparison results in terms of AP and AUC (mean

\pm

standard deviation). Higher values indicate better performance. The best and second-best results are highlighted in bold and underlined, respectively.

Figure 3: Performance comparison on yeast under different selected feature ratios ranging from 2% to 20%.

Dataset	DHLI	EF 2 FS	ENM	GRAFS	I 2 VSLC	MSFS	LLM-Select	FeatureHospital
RL $\downarrow$
SCENE	0.1175 ± 0.0031	0.1552 ± 0.0037	0.1262 ± 0.0045	0.1082 ± 0.0034	0.1085 ± 0.0034	0.1451 ± 0.0260	0.1089 ± 0.0037	0.1113 ± 0.0033
yeast	0.2557 ± 0.0044	0.2639 ± 0.0058	0.2586 ± 0.0074	0.2563 ± 0.0034	0.2539 ± 0.0056	0.2881 ± 0.0074	0.2444 ± 0.0043	0.2501 ± 0.0070
VOC07	0.2796 ± 0.0087	0.3917 ± 0.0212	0.2738 ± 0.0086	0.2798 ± 0.0051	0.2655 ± 0.0035	0.4007 ± 0.0166	0.2567 ± 0.0048	0.2544 ± 0.0059
MIRFlickr	0.1944 ± 0.0047	0.1956 ± 0.0046	0.1689 ± 0.0035	0.1874 ± 0.0030	0.1939 ± 0.0045	0.2567 ± 0.0254	0.1681 ± 0.0029	0.1716 ± 0.0040
mfeat	0.1498 ± 0.0081	0.5131 ± 0.1586	0.0724 ± 0.0264	0.0508 ± 0.0064	0.0377 ± 0.0042	0.1754 ± 0.1026	0.0715 ± 0.0045	0.0283 ± 0.0038

Table 3: Comparison results in terms of RL and ZL (mean

\pm

standard deviation). Lower values indicate better performance. The best and second-best results are highlighted in bold and underlined, respectively. Rankings are determined using the unrounded results.

Dataset	Metric	Full FeatureHospital	Backbone Only	Random Triage	Random Consultation	All Rule-Based Decisions	Single LLM Agent	Without Pharmacist
SCENE	AP $\uparrow$	0.8011	0.7892	0.7956	0.7893	0.7994	0.7963	0.8003
	AUC $\uparrow$	0.6970	0.6784	0.6891	0.6768	0.6954	0.6905	0.6962
	RL $\downarrow$	0.1113	0.1187	0.1145	0.1187	0.1122	0.1142	0.1118
	ZL $\downarrow$	0.9399	0.9543	0.9468	0.9538	0.9400	0.9438	0.9418
mfeat	AP $\uparrow$	0.9360	0.9071	0.8700	0.9217	0.8867	0.9217	0.9235
	AUC $\uparrow$	0.9871	0.9801	0.9617	0.9831	0.9716	0.9830	0.9836

Table 4: Ablation results in terms of AP, AUC, RL, and ZL. The best and second-best results are highlighted in bold and underlined, respectively.

Figure 4: Parameter sensitivity of the objective constructed for yeast.

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Dataset 27B	GPT-5.5 † 27B	Qwen3.5 27B	GPT-5.4 mini	DeepSeek V4-Pro	Claude Sonnet 5	Gemini 3.1 Pro Preview
AP $\uparrow$
SCENE	0.8011 ± 0.0041	0.8039 ± 0.0040	0.7872 $\pm$ 0.0037	0.7875 $\pm$ 0.0041	0.7952 $\pm$ 0.0034	0.7951 $\pm$ 0.0041
yeast	0.6976 ± 0.0081	0.6908 $\pm$ 0.0065	0.6915 $\pm$ 0.0069	0.6989 ± 0.0082	0.6934 $\pm$ 0.0079	0.6910 $\pm$ 0.0074
VOC07	0.6073 $\pm$ 0.0042	0.6064 $\pm$ 0.0042	0.6074 ± 0.0042	0.5964 $\pm$ 0.0050	0.6078 ± 0.0051	0.6068 $\pm$ 0.0043
MIRFlickr	0.7011 $\pm$ 0.0048	0.7008 $\pm$ 0.0046	0.7008 $\pm$ 0.0045	0.6991 $\pm$ 0.0045	0.7019 ± 0.0048	0.7019 ± 0.0044
mfeat	0.9360 ± 0.0056	0.9235 $\pm$ 0.0080	0.9328 $\pm$ 0.0070	0.9213 $\pm$ 0.0098	0.9242 $\pm$ 0.0093	0.9534 ± 0.0054

Appendix

Table 5: Robustness of FeatureHospital across different LLM backbones in terms of AP and AUC (mean

\pm

standard deviation). GPT-5.5 † denotes the reference backbone used in the main experiments. Higher values indicate better performance. The best and second-best results are highlighted in bold and underlined, respectively. Rankings are determined using the unrounded results.

Dataset 27B	GPT-5.5 † 27B	Qwen3.5 27B	GPT-5.4 mini	DeepSeek V4-Pro	Claude Sonnet 5	Gemini 3.1 Pro Preview
RL $\downarrow$
SCENE	0.1113 ± 0.0033	0.1100 ± 0.0028	0.1200 $\pm$ 0.0033	0.1196 $\pm$ 0.0036	0.1155 $\pm$ 0.0033	0.1150 $\pm$ 0.0037
yeast	0.2501 ± 0.0070	0.2591 $\pm$ 0.0060	0.2574 $\pm$ 0.0062	0.2490 ± 0.0063	0.2554 $\pm$ 0.0076	0.2586 $\pm$ 0.0061
VOC07	0.2544 ± 0.0059	0.2552 $\pm$ 0.0058	0.2544 ± 0.0059	0.2672 $\pm$ 0.0067	0.2547 $\pm$ 0.0055	0.2565 $\pm$ 0.0068
MIRFlickr	0.1716 $\pm$ 0.0040	0.1720 $\pm$ 0.0042	0.1716 $\pm$ 0.0040	0.1731 $\pm$ 0.0041	0.1709 ± 0.0040	0.1700 ± 0.0036
mfeat	0.0283 ± 0.0038	0.0353 $\pm$ 0.0064	0.0283 $\pm$ 0.0046	0.0367 $\pm$ 0.0070	0.0340 $\pm$ 0.0054	0.0214 ± 0.0036

Appendix

Table 6: Robustness of FeatureHospital across different LLM backbones in terms of RL and ZL (mean

\pm

standard deviation). GPT-5.5 † denotes the reference backbone used in the main experiments. Lower values indicate better performance. The best and second-best results are highlighted in bold and underlined, respectively. Rankings are determined using the unrounded results.

Model	Calls	Input (K)	Output (K)	Total (K)	Cost (USD)
Qwen3.5-27B	9.0	132.7	21.0	153.8	$0.090
DeepSeek-V4-Pro	9.1	136.7	20.6	157.4	$0.079
GPT-5.4 mini	10.3	159.8	49.0	208.8	$0.340
Gemini 3.1 Pro Preview	8.9	166.0	36.6	202.6	$0.771
Claude Sonnet 5	10.3	243.9	82.1	326.1	$1.309

Appendix

Table 7: Average LLM usage and estimated API cost of one complete FeatureHospital pipeline execution. K denotes one thousand tokens. Costs are calculated using the standard real-time API prices available at the time of evaluation without prompt-caching or batch-processing discounts.

Figure 5: Critical difference diagrams for AP, AUC, RL, and ZL over the seven datasets. Lower average ranks indicate better overall performance. Methods connected by the same thick horizontal line are not significantly different according to the Nemenyi post-hoc test at

\alpha=0.05

(

\mathrm{CD}=3.968

Figure 6: Performance under different feature selection ratios on SCENE and VOC07. Higher values are preferred for AP and AUC, whereas lower values are preferred for RL and ZL.

Figure 7: Performance under different feature selection ratios on MIRFlickr and mfeat. Higher values are preferred for AP and AUC, whereas lower values are preferred for RL and ZL.

Figure 8: Performance under different feature selection ratios on emotions and 3Sources. Higher values are preferred for AP and AUC, whereas lower values are preferred for RL and ZL.

Dataset 27B	Full FeatureHospital	Backbone Only	Random Triage	Random Consultation	Without Pharmacist	All Rule-Based Decisions	Single LLM Agent
AP $\uparrow$
SCENE	0.8011 ± 0.0041	0.7892 $\pm$ 0.0039	0.7956 $\pm$ 0.0062	0.7893 $\pm$ 0.0042	0.8003 ± 0.0042	0.7994 $\pm$ 0.0044	0.7963 $\pm$ 0.0043
yeast	0.6976 $\pm$ 0.0081	0.6995 ± 0.0073	0.6949 $\pm$ 0.0097	0.6910 $\pm$ 0.0084	0.6975 $\pm$ 0.0078	0.6973 $\pm$ 0.0080	0.6977 ± 0.0083
VOC07	0.6073 ± 0.0042	0.6051 $\pm$ 0.0039	0.6019 $\pm$ 0.0094	0.6053 $\pm$ 0.0048	0.6073 ± 0.0045	0.5936 $\pm$ 0.0045	0.6014 $\pm$ 0.0042
MIRFlickr	0.7011 $\pm$ 0.0048	0.7003 $\pm$ 0.0044	0.7015 ± 0.0045	0.6887 $\pm$ 0.0167	0.7011 $\pm$ 0.0047	0.6985 $\pm$ 0.0043	0.7017 ± 0.0041
mfeat	0.9360 ± 0.0056	0.9071 $\pm$ 0.0056	0.8700 $\pm$ 0.0905	0.9217 $\pm$ 0.0213	0.9235 ± 0.0077	0.8867 $\pm$ 0.0056	0.9217 $\pm$ 0.0074

Appendix

Table 8: Ablation results of FeatureHospital in terms of AP and AUC (mean

\pm

standard deviation). Higher values indicate better performance. The Average row reports the arithmetic mean across the seven datasets. The best and second-best results are highlighted in bold and underlined, respectively. Rankings are determined using the unrounded results.

Dataset 27B	Full FeatureHospital	Backbone Only	Random Triage	Random Consultation	Without Pharmacist	All Rule-Based Decisions	Single LLM Agent
RL $\downarrow$
SCENE	0.1113 ± 0.0033	0.1187 $\pm$ 0.0034	0.1145 $\pm$ 0.0046	0.1187 $\pm$ 0.0037	0.1118 ± 0.0036	0.1122 $\pm$ 0.0036	0.1142 $\pm$ 0.0039
yeast	0.2501 $\pm$ 0.0070	0.2492 ± 0.0056	0.2537 $\pm$ 0.0100	0.2573 $\pm$ 0.0087	0.2500 ± 0.0065	0.2515 $\pm$ 0.0068	0.2507 $\pm$ 0.0069
VOC07	0.2544 ± 0.0059	0.2567 $\pm$ 0.0056	0.2609 $\pm$ 0.0116	0.2572 $\pm$ 0.0071	0.2542 ± 0.0062	0.2741 $\pm$ 0.0053	0.2628 $\pm$ 0.0059
MIRFlickr	0.1716 ± 0.0040	0.1724 $\pm$ 0.0041	0.1717 $\pm$ 0.0039	0.1785 $\pm$ 0.0093	0.1716 $\pm$ 0.0040	0.1737 $\pm$ 0.0040	0.1704 ± 0.0040
mfeat	0.0283 ± 0.0038	0.0412 $\pm$ 0.0036	0.0660 $\pm$ 0.0559	0.0351 $\pm$ 0.0117	0.0342 ± 0.0047	0.0535 $\pm$ 0.0031	0.0347 $\pm$ 0.0047

Appendix

Table 9: Ablation results of FeatureHospital in terms of RL and ZL (mean

\pm

standard deviation). Lower values indicate better performance. The Average row reports the arithmetic mean across the seven datasets. The best and second-best results are highlighted in bold and underlined, respectively. Rankings are determined using the unrounded results.

Feature Selection

Feature selection identifies compact, informative variables or input dimensions to improve prediction, efficiency, interpretability, and statistical inference.

See all 198 papers

SEHFS: Structural Entropy-Guided High-Order Correlation Learning for Multi-View Multi-Label Feature Selection
Mar 3, 2026Cheng Peng, Yonghao Li, Wanfu Gao +2Feature Selection Multi-View Learning
the College of Software, Jilin University, Changchun 130012, China · the Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, Changchun 130012, China · the School of Computing and Artificial Intelligence, Southwestern University of Finance and Economics, Chengdu 611130, China +5
FSEVAL: Feature Selection Evaluation Toolbox and Dashboard
Apr 20, 2026Muhammad Rajabinasab, Arthur ZimekFeature Selection Automated Evaluation
Department of Mathematics and Computer Science University of Southern Denmark Odense, Denmark
MOELIGA: a multi-objective evolutionary approach for feature selection with local improvement
Mar 21, 2026Leandro Vignolo, Matias GerardFeature Selection Genetic Algorithms
Research Institute for Signals, Systems and Computational Intelligence (sınc(i)), FICH–UNL/CONICET, Ciudad Universitaria UNL, Santa Fe, Argentina.

Mar 3, 2026J/K move · Enter open · S save

SEHFS: Structural Entropy-Guided High-Order Correlation Learning for Multi-View Multi-Label Feature Selection

Cheng Peng, Yonghao Li, Wanfu Gao, Jie Wen+1

the College of Software, Jilin University, Changchun 130012, China · the Key Laboratory of Symbolic Computation and Knowledge Engineering of the Ministry of Education, Changchun 130012, China · the School of Computing and Artificial Intelligence, Southwestern University of Finance and Economics, Chengdu 611130, China+5

Feature Selection Multi-View Learning

Open paper AskPDF arXiv

In recent years, multi-view multi-label learning (MVML) has attracted extensive attention due to its close alignment to real-world scenarios. Information-theoretic methods have gained prominence for learning nonlinear correlations. However, two key challenges persist: first, features in real-world data commonly exhibit high-order structural correlations, but existing information-theoretic methods struggle to learn such correlations; second, commonly relying on heuristic optimization, information-theoretic methods are prone to converging to local optima. To address these two challenges, we propose a novel method called Structural Entropy Guided High-Order Correlation Learning for Multi-View Multi-Label Feature Selection (SEHFS). The core idea of SEHFS is to convert the feature graph into a structural-entropy-minimizing encoding tree, quantifying the information cost of high-order dependencies and thus learning high-order feature correlations beyond pairwise correlations. Specifically, features exhibiting strong high-order redundancy are grouped into a single cluster within the encoding tree, while inter-cluster feaeture correlations are minimized, thereby eliminating redundancy both within and across clusters. Furthermore, a new framework based on the fusion of information theory and matrix methods is adopted, which learns a shared semantic matrix and view-specific contribution matrices to reconstruct a global view matrix, thereby enhancing the information-theoretic method and balancing the global and local optimization. The ability of structural entropy to learn high-order correlations is theoretically established, and and both experiments on eight datasets from various domains and ablation studies demonstrate that SEHFS achieves superior performance in feature selection.

Abstract

Feature Selection

Explore similar work

SEHFS: Structural Entropy-Guided High-Order Correlation Learning for Multi-View Multi-Label Feature Selection

FSEVAL: Feature Selection Evaluation Toolbox and Dashboard

MOELIGA: a multi-objective evolutionary approach for feature selection with local improvement