Shap
PulseAugur coverage of Shap — every cluster mentioning Shap across labs, papers, and developer communities, ranked by signal.
6 天有情绪数据
-
BOHM method offers zero-cost AI system attribution using routing weights
Researchers have introduced BOHM, a novel method for attributing contributions within compound AI systems that utilize hierarchical routing. Unlike traditional Shapley-based methods, BOHM leverages existing routing weig…
-
ML model struggles with visibility prediction due to data shifts
Researchers have developed a machine learning framework for predicting atmospheric visibility in six South Korean cities, addressing challenges like imbalanced data and distribution shifts. The study employed techniques…
-
Research proves feature ranking impossible under collinearity
A new research paper published on arXiv demonstrates that no feature ranking method can be simultaneously faithful, stable, and complete when features are collinear. The study proves this impossibility and quantifies it…
-
New metric quantifies AI explanation fragility in cybersecurity
This paper introduces a novel metric, the Explanability Fragility Score, to quantify instability in AI explanations within cybersecurity intrusion detection systems. The research demonstrates that multicollinearity, a s…
-
SHAP guide details ML model interpretability workflows
This guide provides a practical framework for interpreting machine learning models using SHAP explainability workflows. It details how to train tree-based models and compares various SHAP explainers, such as Tree, Exact…
-
New methods enhance AI model explainability for images and tabular data
Researchers have developed two new methods for improving feature attribution in machine learning models. Spectral Integrated Gradients (SIG) uses singular value decomposition to create attribution paths that progress fr…
-
New RoSHAP metric enhances stable feature attribution in ML
Researchers have introduced RoSHAP, a new framework and metric designed to improve the stability and interpretability of feature attribution in machine learning models. Traditional attribution methods often yield incons…
-
Machine learning framework aids diabetes detection and subtype analysis
Researchers have developed a novel three-stage machine learning framework to address the complexities of diabetes management. The first stage benchmarks various classifiers for detecting diabetes and identifies key pred…
-
CatNet paper introduces SHAP for feature importance in LSTM FDR control
Researchers have introduced CatNet, a novel algorithm designed to control the False Discovery Rate (FDR) and identify significant features within Long Short-Term Memory (LSTM) networks. This method utilizes the derivati…
-
New algorithm computes exact Shapley values for product-kernel methods
Researchers have developed PKeX-Shapley, a novel algorithm designed to compute exact Shapley values for product-kernel methods in machine learning. This new method leverages the multiplicative structure of product kerne…
-
Neural-Actuarial Longevity Forecasting: Anchoring LSTMs for Explainable Risk Management
Researchers have developed a new neural-actuarial framework called Hybrid-Lift to improve longevity forecasting. This approach combines Hierarchical LSTM networks with a Mean-Bias Correction anchoring mechanism to addre…
-
GRALIS framework unifies linear attribution methods for deep neural networks
Researchers have introduced GRALIS, a novel mathematical framework designed to unify various linear attribution methods used in Explainable AI (XAI). This framework establishes a canonical representation for attribution…
-
ML models show inconsistent feature importance in electrospinning research
A new research paper explores the consistency of feature importance across various machine learning models in the context of electrospinning. The study evaluated 21 different ML models using SHAP values to assess the re…
-
AI decodes driver behavior and auditory signals using advanced machine learning
Researchers have developed a new framework for classifying driver behavior using a combination of physiological signals like EEG, EMG, and GSR. The system employs SHAP-based feature selection to identify the most predic…
-
New phi-table method enhances global SHAP explanations for tabular models
Researchers have introduced the $\phi$-table, a new method for statistically explaining global SHAP values in tabular black-box regression models. This approach moves beyond simple feature importance rankings to provide…
-
Researchers develop stable, explainable AI for elderly fall detection
Researchers have developed a new framework for skeleton-based fall detection that uses a temporally stabilized attribution mechanism called T-SHAP. This method enhances the interpretability of AI models used in elderly …
-
New neural network architectures offer aligned explanations for AI predictions
Researchers have introduced Pointwise-interpretable Networks (PiNets), a novel architecture designed to ensure that explanations for neural network predictions genuinely reflect the model's reasoning process. These netw…
-
AI research in 2026: Spiking networks, web search agents, and Anthropic's dangerous Claude Mythos
New research indicates that Binary Spiking Neural Networks can serve as reliable causal models, outperforming existing methods like SHAP in explaining AI decisions. Separately, a novel bi-level multi-agent system called…
-
GRASP framework enhances medical prediction with robust feature selection
Researchers have developed GRASP, a new framework for feature selection in medical prediction tasks. GRASP combines Shapley value attributions with group $L_{21}$ regularization to identify compact and interpretable fea…
-
AI framework cuts brain microstructure scan time by half
Researchers have developed a new, faster protocol for quantifying human gray matter microstructure using diffusion MRI. By employing an Explainable AI (XAI) framework, specifically XGBoost and SHAP, they identified an o…