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 28 for “"Dataset bias"”.
-
Training Models to Ignore Dataset Bias
… correlations that were introduced during dataset creation, and are not fundamental to the target task. For example, sentence entailment datasets can have spurious word-class correlations if nearly all contradiction sentences contain the word ``not'', and image recognition datasets can have …
-
Addressing Dataset Bias in Android Malware Classifiers under Distribution Shift
… the causes for such drift seen in popular datasets are less explored. This thesis focuses exploring current Android malware datasets and their validity under real-world evaluation conditions. First, we investigate biases in datasets that attempt to capture this drift, and connecting them to …
-
Transfer Learning, Model Interpretation, and Dataset Bias Analysis for Automated Violence Detection from Video
… evaluation using a benchmark violence detection dataset, however, these methods cannot be shown to improve final model accuracy. The final transfer learning and hyper-parameter tuning experiments of this thesis remove these new techniques, and the model achieves reasonably competitive accuracy on …
-
Improving cross-dataset generalization in image classification with contrastive representation learning
… in other acquisition parameters, leads to datasets with idiosyncratic footprints and, therefore, limited generalization of the trained deep learning model to other sets of images different from the training set. As a result, more work is required toward improving the cross-dataset …
-
Advancing Multi-Modal Deep Learning: Towards Language-Grounded Visual Understanding
… artificial intelligence. Recently, multiple datasets and algorithms have been created as proxy tasks towards this goal, with visual question answering (VQA) being the most widely studied. In VQA, an algorithm needs to produce an answer to a natural language question about an image. However, …
-
Understanding Visual Information: from Unsupervised Discovery to Minimal Effort Domain Adaptation
… resulting in the generation of very large scale datasets. This has lead to remarkable performance advances in tasks such as scene recognition or object detection, however, at a considerable high cost in terms of human labeling effort. In light of the labeling cost issue, together with the dataset …
-
An offline writer independent signature verification method with robustness against scalings and rotations
… techniques on Western-style signatures (CEDAR dataset), while also obtaining competitive results on South Asian-style handwriting (Bangla and Hindi datasets). Furthermore, a wide range of experiments demonstrate that VerSig-R is the most robust in relation to differences in scale and rotation …
-
RNA secondary structure prediction using hybrid methods
… high accuracy on the human-transcriptome (HT) dataset. However, challenges in performance on the bpRNA dataset are observed, potentially stemming from dataset bias and limitations of the S1 model. Nevertheless, the models exhibit strong generalization capabilities compared to other deep …
-
Learning from positive and unlabeled data for imbalanced datasets: theory and real-world applications
… approach. ML algorithms analyze large datasets to uncover patterns and make predictions, improving with more data. This adaptive learning process en- ables applications in image and speech recognition, natural language processing, and predictive analytics, handling tasks that are …
-
Enhancing Shear Design for Prestressed Concrete Beams Through Reliability-Based Data-Driven Framework
… integrating physical constraints, correcting for dataset bias using Finite Element Analysis (FEA) augmentation, and incorporating the error norm of the Reserve Shear Strength (RSS) index directly into the optimization process to ensure physically consistent and robust predictions. Finally, the …
-
Enhancing Shear Design for Prestressed Concrete Beams Through Reliability-Based Data-Driven Framework
… integrating physical constraints, correcting for dataset bias using Finite Element Analysis (FEA) augmentation, and incorporating the error norm of the Reserve Shear Strength (RSS) index directly into the optimization process to ensure physically consistent and robust predictions. Finally, the …
-
Enhancing Shear Design for Prestressed Concrete Beams Through Reliability-Based Data-Driven Framework
… integrating physical constraints, correcting for dataset bias using Finite Element Analysis (FEA) augmentation, and incorporating the error norm of the Reserve Shear Strength (RSS) index directly into the optimization process to ensure physically consistent and robust predictions. Finally, the …
-
Enhancing Shear Design for Prestressed Concrete Beams Through Reliability-Based Data-Driven Framework
… integrating physical constraints, correcting for dataset bias using Finite Element Analysis (FEA) augmentation, and incorporating the error norm of the Reserve Shear Strength (RSS) index directly into the optimization process to ensure physically consistent and robust predictions. Finally, the …
-
Multi-Modal Data Fusion, Image Segmentation, and Object Identification using Unsupervised Machine Learning: Conception, Validation, Applications, and a Basis for Multi-Modal Object Detection and Tracking
… While the increase in instruments, and therefore datasets, is a boon for those aiming to study the complexities of the various Earth systems, it can also present a large number of new challenges. With this information in mind, our group has set our aims on combining datasets with different spatial …
-
Fine-tuning foundation models for activity of daily living analysis
… methodology involves pretraining on extensive datasets and subsequently fine-tuning on smaller, more specialized datasets. This process allows the model to maintain its general representation capability while improving its adaptability to specific tasks like activity classification, people …
-
Fairness in Machine Learning Methods for Surgical Skill Assessment
… ML models in this context is the potential for bias, which can stem from various sources, including the data. Such biases can skew results, leading to unfair assessments and potentially impacting surgeon careers and patient outcomes. Utilizing a pairwise comparison framework, we introduce bias …
-
ENHANCING TRUST IN SAFETY-CRITICAL AI: THE ROLE OF CONTEXTUAL INTEGRITY IN SPACE SYSTEMS CYBERSECURITY AND RADIOMICS
… domains, yet failures caused by domain shift, dataset bias, and insufficient contextual awareness continue to thwart reliability and trust. This dissertation advances the thesis that contextually informative input data is a core safety principle, enabling robust and generalizable AI systems. …
-
Socially Responsible and Factual Reasoning for Equitable AI Systems
… to measure intent, factuality and social bias of language. We first introduce reaction frames, which allow us to distill knowledge of cognitive and physical effects on readers like implied actions (e.g. given the false statement ``Water boiled with garlic cures coronavirus,'' we can infer …
-
LEARNING FROM INCOMPLETE AND HETEROGENEOUS DATA
… unknown. With experiments on multiple benchmark datasets we show that the proposed recognition systems are able to perform better compared to existing approaches. Furthermore, we also recognize that in many real world conditions training and testing data distributions are often different. Due to …
Page 1 of 2