{"slug": "predicting-privacy-leakage-from-weight-spectral-density", "title": "Predicting Privacy Leakage from Weight Spectral Density", "summary": "A new arXiv paper (2609.11780v1) reports that spectral metrics from the WeightWatcher framework can predict membership inference attack (MIA) vulnerability in machine learning models without training expensive shadow models. Evaluating image and tabular classification tasks, the authors found stable rank shows a strong positive correlation with overall MIA success, while Log alpha-Norm shows a consistent negative correlation with MIA vulnerability at the low false-positive regime, associations stronger than those from the generalisation gap. The authors conclude that neural network spectra may contain privacy-leakage information not captured by conventional overfitting measures, motivating spectral analysis for scalable privacy auditing.", "body_md": "arXiv:2609.11780v1 Announce Type: new \nAbstract: Membership inference attacks (MIAs) are widely used to audit the privacy disclosure risk of machine learning models, however current state-of-the-art attacks require training computationally expensive shadow models, making large-scale privacy evaluation impractical. In this work, we investigate whether inexpensive spectral metrics derived from the heavy-tailed self-regularisation framework can serve as proxies for MIA vulnerability. We evaluate several WeightWatcher spectral metrics on image and tabular classification tasks and compare their relationship with MIA privacy leakage against conventional measures of generalisation. Across datasets, stable rank exhibits a strong positive correlation with overall MIA success, while Log alpha-Norm shows a consistent negative correlation with MIA vulnerability at the low false-positive regime. These associations are observed to be stronger than those obtained using the generalisation gap. The results indicate that neural network spectra may contain information about privacy leakage that is not fully captured by conventional measures of overfitting, motivating spectral analysis as a promising direction for scalable privacy auditing.", "url": "https://wpnews.pro/news/predicting-privacy-leakage-from-weight-spectral-density", "canonical_source": "https://www.machinebrief.com/news/predicting-privacy-leakage-from-weight-spectral-density-icv6", "published_at": "2026-09-11 04:00:00+00:00", "updated_at": "2026-09-11 05:27:18.067312+00:00", "lang": "en", "topics": ["machine-learning", "ai-safety", "ai-research", "ai-ethics"], "entities": ["arXiv", "WeightWatcher", "membership inference attacks", "stable rank", "Log alpha-Norm"], "alternates": {"html": "https://wpnews.pro/news/predicting-privacy-leakage-from-weight-spectral-density", "markdown": "https://wpnews.pro/news/predicting-privacy-leakage-from-weight-spectral-density.md", "text": "https://wpnews.pro/news/predicting-privacy-leakage-from-weight-spectral-density.txt", "jsonld": "https://wpnews.pro/news/predicting-privacy-leakage-from-weight-spectral-density.jsonld"}}