Research
Detecting Unreliable Classifications from Quantized Models
arXiv:2610.18575v1 Announce Type: new Abstract: Post-training INT8 quantization is typically validated by checking accuracy retention, which requires ground-truth labels often unavailable in low-resou
arXiv:2610.18575v1 Announce Type: new Abstract: Post-training INT8 quantization is typically validated by checking accuracy retention, which requires ground-truth labels often unavailable in low-resource clinical settings. We show a label-free alternative: comparing a model's own output before and after quantization. We compute cosine similarity between the FP32 and INT8 models' output probability vectors for each prediction, which requires only a single forward pass. This output-agreement similarity separates lower-accuracy predictions across four evaluated architectures (ResNet-50, VGG-16-BN, Wide ResNet-50, and ViT-B/16) under fixed-percentile splits, outperforming a confidence-only baseline for all four. These results show that quantized models can be compared to their baselines to assess their drift and determine whether they are well suited for clinical deployment.
Related
- SpatialUQ: Post-Hoc Uncertainty Quantification from Spatial Consistency in Black-Box Vision Models
- SubRot: Signed Gradient Subspace Calibration for VLM Rotation Quantization
- CAR-SAM: Cross-Attention Reconstruction for Post-Training Quantization of the Segment Anything Model
Source: arXiv cs.CV | 2026-10-14