Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
ISBN 9781510838819
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 8roles
background 2polarities
background 2representative citing papers
Tensor similarity is a symmetry-invariant metric that measures functional equivalence between tensor-based networks using a recursive algorithm for cross-layer mechanisms.
Cross-sample prediction churn between bootstrap-trained classifiers reaches 8-22% on chemistry benchmarks; K-bootstrap bagging reduces it 40-54% and twin-bootstrap with sym-KL consistency loss reduces it a further median 45% at matched 2x compute.
LasRepair++ pairs an LLM instructor with an SLM corrector, refines context via EM, and down-weights uncertain repairs using column-calibrated confidence, reporting 18.1% average F1 gain over baselines on data repair tasks.
Circuit-based metrics from Vision Transformer internals provide better label-free proxies for generalization under distribution shift than existing methods like model confidence.
Proposes the REAL framework for ML systems requirements engineering that weaves data/model/system requirements, uses failure-driven exploration, and supports iterative traceable refinement, shown via an autonomous driving example.
Solution concentration is the only robust feature across ML models for electrospinning while flow rate and applied voltage show high model-dependent variability in importance rankings.
Proposes a method to find multiple similar-performing models with distinct context-aware characteristics on the METABRIC dataset.
citing papers explorer
-
Toward Calibrated, Fair, and accurate Deepfake Detection
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
-
When Are Two Networks the Same? Tensor Similarity for Mechanistic Interpretability
Tensor similarity is a symmetry-invariant metric that measures functional equivalence between tensor-based networks using a recursive algorithm for cross-layer mechanisms.
-
Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings
Circuit-based metrics from Vision Transformer internals provide better label-free proxies for generalization under distribution shift than existing methods like model confidence.
-
Cross-Model Consistency of Feature Importance in Electrospinning: Separating Robust from Model-Dependent Features
Solution concentration is the only robust feature across ML models for electrospinning while flow rate and applied voltage show high model-dependent variability in importance rankings.
-
Finding Multiple Interpretations in Datasets
Proposes a method to find multiple similar-performing models with distinct context-aware characteristics on the METABRIC dataset.