Exploring 3D Dataset Pruning

Xiaohan Zhao; Xinyi Shang; Jiacheng Liu; Zhiqiang Shen

cs.CVFeb 28, 2026

Exploring 3D Dataset Pruning

Authors: Xiaohan Zhao, Xinyi Shang, Jiacheng Liu, Zhiqiang Shen

Organizations: VILA Lab, Department of Machine Learning, MBZUAI.

Abstract

Dataset pruning has been widely studied for 2D images to remove redundancy and accelerate training, while particular pruning methods for 3D data remain largely unexplored. In this work, we study dataset pruning for 3D data, where its observed common long-tail class distribution nature make optimization under conventional evaluation metrics Overall Accuracy (OA) and Mean Accuracy (mAcc) inherently conflicting, and further make pruning particularly challenging. To address this, we formulate pruning as approximating the full-data expected risk with a weighted subset, which reveals two key errors: coverage error from insufficient representativeness and prior-mismatch bias from inconsistency between subset-induced class weights and target metrics. We propose representation-aware subset selection with per-class retention quotas for long-tail coverage, and prior-invariant teacher supervision using calibrated soft labels and embedding-geometry distillation. The retention quota also serves as a switch to control the OA-mAcc trade-off. Extensive experiments on 3D datasets show that our method can improve both metrics across multiple settings while adapting to different downstream preferences. Our code is available at https://github.com/XiaohanZhao123/3D-Dataset-Pruning.

Figures & tables

Figure 1

Figure 2: Illustration of our 3D-Pruner framework, which comprises: (1) base principles (utilizing embedding signals, minimum floor selection, and optimizing structural likelihood in post-pruning training) that remain robust and beneficial across different priors, derived from theoretical analysis of the shared region, and (2) a steering warper that balances between the two priors.

Figure 3: Selection composition across algorithms. Incomparable scalar scores (loss, EL2N) bias selection toward many-shot classes. Geometric embedding selection is more robust but still partially affected by class distribution, under-selecting certain classes. SGS mitigates this, also enabling a flexible balancing.

	Hard		KD
Rebalancing	OA	mAcc	OA	mAcc
CB-Samp. (CB)	79.02	61.20	82.09	64.56
CB-Loss	77.45	62.16	81.94	63.57
IB-Samp. (IB)	79.92	58.41	82.04	64.02
Sqrt-Samp. (Sqrt)	80.00	60.10	81.21	64.81
Mean $\pm$ Std	79.10 $\pm$ 1.18	60.47 $\pm$ 1.61	81.82 $\pm$ 0.41	64.24 $\pm$ 0.56

Table 2: KD improves accuracy and reduces sensitivity to rebalancing. Rebalancing policies include instance-balanced sampling (IB), class-balanced sampling (CB), and square-root sampling (Sqrt) ( Kang et al., 2019 ) , plus class-balanced loss (CB-Loss) ( Cui et al., 2019 ) . Results are obtained on ShapeNet55 with PointNet++, with imbalanced set selected by FL-RBF ( Wei et al., 2015 ) .

Signal	$\rho$ (Corr. w/ Size) $\downarrow$	$R^{2}$ (Dep. on Size) $\downarrow$	Overlap $\uparrow$	Imbalance Ratio $\downarrow$
Embedding	-0.306	0.148	96%	1.88x
Loss	0.635	0.145	26%	40.94x
EL2N	0.594	0.164	30%	33.01x
Entropy	0.731	0.481	26%	7.84x

Table 3: Signal Audit. We measure correlation (

\rho

) and dependence (

R^{2}

) of class-aggregated signal magnitude on class size, and head/tail score overlap. For embeddings, magnitude is distance to the class center. Embedding signals show minimal class-frequency dependence and high cross-class overlap , whereas scalar signals (Loss, EL2N) are dominated by the training prior, causing extreme selection imbalance (up to 40

\times

Panel of Figure 4 — Figure 4: Comparison of pruning methods across various datasets, models, and budgets on point cloud modality.

Panel of Figure 5 — Figure 5: Impact of robust distillation methods: Calibrated Soft Label (CSL) and Embedding Geometry Distillation (EGD) on OA and mAcc. Results are obtained on PointNet++.

Figure 6: Comparison of pruning methods on mesh modality, legend following Fig. 4

		ModelNet40						ScanObjectNN						ShapeNet55
		$m=400$		$m=800$		$m=1600$		$m=150$		$m=300$		$m=600$		$m=550$		$m=1100$		$m=2200$
Metric Type	Selection Policy	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc
Scalar Scores	Global Selection
	Loss	56.36	47.76	75.97	65.97	88.24	82.43	35.84	35.60	52.36	51.37	63.15	61.10	55.21	20.12	60.13	25.34	67.87	25.38
	GradNorm	44.61	35.72	70.22	59.32	87.96	79.62	45.70	36.22	51.84	42.47	60.40	50.63	62.82	22.60	67.39	27.18	71.06	31.54
	Entropy	58.71	49.24	77.84	68.37	88.37	82.74	34.46	34.80	50.97	49.68	63.21	62.30	66.47	24.06	75.59	33.00	79.57	41.27

Table 4: Fair comparison of selection strategies on ModelNet40 , ScanObjectNN , and ShapeNet55 with Calibrated Soft Label and Embedding Geometry Distillation from 3D-Pruner applied to all baselines.

Figure 7: Preferences steering with different K.

Figure 8: Transferability of 3D-Pruner across different architecture. Results are based on $m=1100$ on ShapeNet55.

Appendix figures & tables4 assets

Supplementary material from the paper’s appendix.

Appendix

	ShapeNet55		ModelNet40		ScanObjectNN
Model	OA	mAcc	OA	mAcc	OA	mAcc
PointNeXt-S	91.11	83.47	93.15	90.60	87.20	85.80
PointNet++	91.08	82.83	92.95	90.34	86.12	84.37
PointVector-S	91.00	82.96	90.15	85.31	87.37	85.64

Appendix

Table 5: Cross-dataset evaluation across 3D backbones. Overall accuracy (OA) and mean accuracy (mAcc) on ShapeNet55, ModelNet40, and ScanObjectNN.

Panel of Figure 9 — Figure 9: Additional results on relaxed budgets on ShapeNet55. The gap on OA becomes smaller given larger budget, yet our 3D-Pruner remains the best on the mAcc.

		ModelNet40
		$m=400$		$m=800$		$m=1200$
Metric Type	Selection Policy	OA	mAcc	OA	mAcc	OA	mAcc
Scalar Scores	Stratified Sampling
	Balance-EL2N	60.60	56.71	76.83	74.72	82.85	80.68
	NUCS-EL2N	57.06	52.99	72.89	70.68	81.27	78.40
	CCS-CP-EL2N	68.25	57.17	78.83	70.07	83.12	75.21

Appendix

Table 6: Comparison of different pruning methods on ModelNet40, MeshMAE. The pruning methods are selected from the most competitive ones based on the performance on the MeshNet.

	PointNeXt-S						PointVector-S
	$m=550$		$m=1100$		$m=2200$		$m=550$		$m=1100$		$m=2200$
Method	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc	OA	mAcc
Loss	71.32	31.82	75.90	41.46	78.38	40.01	71.75	28.02	74.71	34.17	76.35	35.37
GradNorm	76.75	37.96	79.36	43.69	82.25	53.49	80.25	49.30	82.76	56.60	84.31	60.24
Entropy	72.01	30.96	74.98	34.71	78.91	41.62	71.99	31.82	75.93	38.64	78.05	42.24
EL2N	71.71	31.69	76.57	36.99	78.04	41.45	70.47	26.15	74.42	30.29	78.29	37.66

Appendix

Table 7: Additional results on PointNeXt and PointVector.

Dataset Pruning

Dataset pruning selects compact, informative training subsets to reduce computation while preserving or improving model performance.

See all 22 papers

Dataset Pruning from First Principles: A Label-Free Linear Programming Approach
Oct 7, 2026Rodrigo Schuller, Francisco GanacimDataset Pruning Training Data Selection
Instituto de Matem´atica Pura e Aplicada (IMPA) Rio de Janeiro, Brazil
Data Pruning: Redundant, Problematic, and Interdependent Samples
Jun 20, 2026Leon Freese, Marthinus W. TheunissenDataset Pruning Training Data Selection
Faculty of Engineering, North-West University, South Africa · Centre for Artificial Intelligence Research (CAIR), South Africa · National Institute for Theoretical and Computational Sciences (NITheCS), South Africa
Label-Efficient Dataset Pruning via Semi-Supervised Pseudo-Labeling
May 22, 2026Yeseul Cho, Baekrok Shin, Changmin Kang +1Dataset Pruning Coreset Selection
Kim Jaechul Graduate School of AI, KAIST

Oct 7, 2026J/K move · Enter open · S save

Dataset Pruning from First Principles: A Label-Free Linear Programming Approach

Rodrigo Schuller, Francisco Ganacim

Instituto de Matem´atica Pura e Aplicada (IMPA) Rio de Janeiro, Brazil

Dataset Pruning Training Data Selection Variance Reduction

Open paper AskPDF arXiv

Dataset pruning reduces a large training set to a representative subset while preserving model performance. Existing geometry-based methods typically assume that nearby points in embedding space share similar properties. Rather than imposing this assumption, we derive geometric selection criteria by reformulating unbiased subset selection as a variance minimization problem. Unbiasedness ensures that unweighted subset averages recover full-dataset averages in expectation, including losses and gradients at fixed model parameters. Specifically, we characterize a family of unbiased subset selection algorithms as a high-dimensional polytope. In this context, minimizing the expected sampling variance is a linear objective. Differences in sampling variance, averaged over rigid motions, admit closed-form pairwise expressions. Because the polytope has high dimension, directly applying standard linear programming is impractical. We instead use these expressions to construct an efficient vertex walk that optimizes an approximation of the variance objective while preserving unbiasedness, yielding a method that requires neither labels nor model training during selection. Across CIFAR-10, MNIST, and CelebA benchmarks, our method matches or exceeds uniform sampling in mean test accuracy at every evaluated budget and outperforms competing geometric methods in several settings, particularly at small selection budgets. Beyond dataset pruning, the same framework reduces stochastic-gradient variance by increasing diversity within mini-batches while keeping the batch size unchanged.

Abstract

Dataset Pruning

Explore similar work

Dataset Pruning from First Principles: A Label-Free Linear Programming Approach

Data Pruning: Redundant, Problematic, and Interdependent Samples

Label-Efficient Dataset Pruning via Semi-Supervised Pseudo-Labeling