Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 37 for “"Feature attribution"”.
-
Proxy Gradient Estimation for Black-Box Feature Attribution
Feature attribution is a widely adopted form of explanation that reveals the decision-making process of data-driven models. Its objective is to quantify the contribution of input components to a certain model outcome, highlighting the most prominent features and thereby providing an intuitive …
-
The Weighted Möbius Score: A Unified Framework for Feature Attribution
Feature attribution aims to explain the reasoning behind a black-box model's prediction by identifying the impact of each feature on the prediction. Recent work has extended feature attribution to interactions between multiple features. However, the lack of a unified framework has led to a …
-
Martingale Methods for Structural Change Detection and Feature Attribution in Dynamic Networks
… without distributional assumptions, exact feature attribution, and early detection with rigorous false alarm control. We construct additive martingale statistics from multiple graph features using conformal prediction, providing false alarm guarantees via Ville's inequality. Our key …
-
Understanding Biomedical Machine Learning Models
… by these algorithms has grown correspondingly. Feature attribution methods are one popular approach to explain these black box models, but are limited in their expressive capacity. We therefore propose three approaches to go beyond the shortcomings of existing feature attribution methods. The …
-
Model Interpretability for Natural Language Processing Applications
… literature for extracting explana- tions (e.g. feature attribution methods), some faced criticism about how faith- ful they are. Furthermore, explanation faithfulness also depends on the model employed, where highly parametrised models have been shown to produce less faithful explanations. …
-
Enhancing deep learning security through explainability and robustness
… neural networks in production using feature attribution methods, and introduces a new framework for evaluating the robustness of attribution methods. Firstly, TrojDRL is introduced as a tool for exploring and evaluating backdoor attacks on deep reinforcement learning agents. TrojDRL …
-
Explainable Machine Learning and Applications in Protein-Ligand Complex Structure Prediction
… and structures. The first portion deals with feature attribution techniques, which attribute attribution scores on a per-instance basis to a machine learning model that represent that model locally around that instance as linear. Three methods are proposed: expected gradients, attribution …
-
Application of Explainable Graph Neural Networks in Drug and Biomarker Discovery for Human Diseases
… improves predictive performance and stabilizes feature attribution, enabling more reliable identification of key molecular moieties in lead optimization. Second, this dissertation presents Graph Artificial Intelligence for Pediatric Oncology (GAIPO), a scalable data and model framework that …
-
G2G - From Data Generation to Model Generalisation
… to reveal interpretable patterns and improve feature attribution; generating synthetic examples to support few shot classification; extracting structure from unlabelled inputs through self supervised objectives; and creating artificial patterns to guide visual reidentification of animals in …
-
Exploring protein-protein interactions using high-throughput datasets and deep learning
… (3) training models on limited structural features for PPI prediction for specific families, and (3) applying feature attribution techniques to interpret sequence-to-function models. Together, this work supports the continued development of experimental and computational tools to deepen our …
-
Model monitoring in the absence of labeled data via feature attributions distributions
… that unifies all its sections. It explores feature attribution distributions for both monitoring dimensions. Using these feature attribution explanations, we can exploit their theoretical properties to derive and establish certain guarantees and insights into model monitoring.<br/><br/>For …
-
Towards Trustworthy Speech Technologies: Enhancing Explainability and Mitigating Gender Bias in Speech Transcription and Translation
… in ASR and AST systems. It introduces the first feature attribution technique for modern autoregressive S2T models, providing fine-grained explanations for ASR and AST outputs by assessing the relevance of audio features and previously generated tokens. To investigate factors contributing to …
-
Computational and Data-Driven Design of Perturbed Metal Sites for Catalytic Transformations
… theories with graph neural network and attribution analysis enables accurate prediction equipped with mechanistic insights. In addition, novel physics effect controlling surface reactivity beyond conventional understanding is uncovered. Secondly, we develop a computational and …
-
Scientific machine learning for astrophysical and cosmological inference.
… are systematically compared using feature-attribution analyses. Second, image-based glitch classification is investigated using convolutional neural networks and Vision Transformers applied to time–frequency spectrograms, with emphasis on robustness under controlled distribution …
-
Towards Reliable and Trustworthy Deep Learning through Explainability and Interpretability
… new post-hoc explanation methods that improve feature attribution and concept-based explanations. These methods provide faithful decision cues by modeling meaningful feature interactions and extracting faithful coherent concepts, enabling more reliable understanding of why a model predicts a …
-
Decoding Lassi Fermentation Using Experimental and Machine Learning Approaches: Longitudinal Analysis of Microbial Kinetics and Physicochemical Dynamics
… demonstrated strong generalization, and feature attribution analyses identified storage time as the dominant global predictor, with formulation acting as an offset modulator of trajectory magnitude. These results demonstrate that lassi fermentation exhibits nonlinear, …
-
Trustworthy Machine Learning: Specification, Verification, and Explanation
… certified guarantees on the robustness of feature attribution methods. Building on this, our approach extends to the Stability Certification Algorithm (SCA), offering more flexible and practical guarantees. Together, these methods provide non-trivial, practical, and model-agnostic …
-
Concept drift–resilient malware analysis in evolving threat environments
… through Integrated Gradients--based feature attribution. Behavioural analysis is performed using capabilities identified by the CAPA tool and techniques from the MITRE ATT&CK framework. The results show that detected drift is often associated with identifiable feature changes and, in …
-
Extracting and Harnessing Interpretation in Data Mining
… goal of interpretation is providing the attribution of input features to model outputs. To adapt feature attribution to explaining outlier detection, I propose Contextual Outlier Interpretation (COIN). Second, to overcome the limitation of attribution methods that do not explain internal …
-
Towards Explainable Time Series: An Exploration Of Post-hoc Local Explanations From A Multi-disciplinary Perspective
… We adapt both LIME and SHAP to provide two novel attribution methods which are specifically designed for univariate and multivariate time series. <br/><br/>The Shapley value is one of the most popular methods for value attribution. However, when applied for feature attribution it has been known to …
Page 1 of 2