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 32 for “"self training"”.
-
Self-Training for Natural Language Processing
… and explainability. In this thesis, we explore self-training methods for mitigating the data distribution gaps between training and evaluation domains and tasks. In contrast to traditional self-training methods that study the best practice of training models with real data and pseudo labels, we …
-
Self-Training and Calibration for Learning with Limited Data
Semi-supervised learning methods such as self-training are able to leverage unlabeled data, which is widely available, as opposed to only using labeled data like many successful supervised learning methods. One part of self-training is to use a trained model to create pseudo-labels for unlabeled …
-
Self-training artificial neural networks for risk reduction in nuclear power operations
The risk reduction potential of the class of artificial neural networks based on the Barto-Sutton architecture is established. The risk associated with nuclear power operations is characterized by sequences of discrete events, such as technical specification violation. The Barto-Sutton architecture …
-
Self-training for cyberbully detection: Achieving high accuracy with a balanced multi-class dataset
… of a balanced dataset specifically designed for training the ML/ DL models. To overcome the challenge of limited labeled data, we employ a semi-supervised self-training algorithm, which effectively expands the size of the labeled dataset. By leveraging real-world social media data, we train and …
-
Harnessing the Power of Self-Training for Gaze Point Estimation in Dual Camera Transportation Datasets
… limitations, this thesis investigates the use of self-learning techniques such as semi-supervised learning and self-training, which can reduce the need for labeled data while maintaining high accuracy. The proposed method is evaluated on the DGAZE dataset and achieves a 57.2\% improvement in …
-
Weakly supervised aspect extraction for domain-specific texts
… model is equipped with multi-head attention and self-training. The multi-head attention is learned from the seed words to ensure that the aspect-related words in text segments are weighted higher than those unrelated ones. The self-training mechanism provides more pseudo labels in addition to …
-
Teaching Graphing Using Enhanced Written Instructions: Does Chunk Size Matter
… can be taught using various methods, perhaps self-training methods could prove both effective and efficient due to the self-guided nature of the methods. One effective self-training method for graphing is enhanced written instructions (EWI). While the literature has demonstrated EWI’s …
-
A Study of the Technology Competencies of Preservice Secondary Mathematics Teachers
… prepared to teach with technology with some more self-training. However, their views about the role that technology should play in the classroom indicated that full appreciation of technology's role in reforming education needs more than the formal integration of technology into a preservice …
-
Transfer Learning For Spoken Language Processing
… the ASR performance drops considerably when the training data distribution does not match the distribution that the model encounters during deployment (target domain). A straightforward remedy is collecting labeled data in the target domain and re-training the source domain ASR model. However, it …
-
Learning without Labels - Reducing Supervision in Training, Inference, and Evaluation of Deep Neural Networks
… across the entire deep learning pipeline. In the training phase, we explore unsupervised fine-tuning, focusing on Source-Free Unsupervised Domain Adaptation scenarios in visual tasks such as Facial Expression Recognition and video-based Action Recognition, primarily leveraging self-supervision and …
-
Unsupervised Domain Adaptation per la rilevazione di oggetti e riconoscimento di azioni
… innovative basate su adversarial learning, self-training e image-to-image translation per apprendere rappresentazioni invarianti rispetto al dominio che possono generalizzare su singoli o multipli domini di destinazione. Per il riconoscimento delle azioni, analizziamo la capacità dei metodi …
-
Weakly-supervised text classification
… classification models suffer from the lack of training data in many real-world applications. Although many semi-supervised and weakly-supervised text classification models exist, they cannot be easily applied to deep neural models and meanwhile support limited supervision types. In this work, …
-
Mining social media stimulus from news article text using weakly-supervised narrative classification
… challenges we need to solve: 1) Lack of training data: the given news article data does not have labeling for narratives and we can not afford manual labeling other than a small evaluation set. 2) The complexity in narratives: narratives are defined in a more complex way comparing to the …
-
Social media adoption by microbusinesses
… to address concerns, along with quick start and self-training, helped to adopt social media. Participants needed to focus on concrete experience, work-place learning and personal knowledge for learning to use social media. Usefulness arising from improved communication, fitness and medium …
-
Multimodal Representation Learning for Agentic AI Systems
… methods. Our approach utilizes progressive self-distillation and soft image-text alignments to model the many-to-many correspondences found in noisy web-harvested datasets. Extensive evaluation demonstrates that our method consistently outperforms CLIP across multiple benchmarks, including …
-
Feature extractor stacking for cross-domain few-shot learning
… independently, use cross-validation to extract training data for stacked generalisation from the support set, and learn a simple linear stacking classifier from this data. We evaluate our FES methods on the well-known Meta-Dataset benchmark, targeting image classification with convolutional …
-
Technology in Mathematics Education: A Descriptive Study of the Availability and Uses of Calculators and Computers in Public High School Mathematics Classes in the State of Virginia
… the State of Virginia through the use of a self-administered mail questionnaire. From these questionnaires, the data gathered about calculator and computer availability, factors influencing teachers' professional development, and actual usage in SOL courses were analyzed to provide a picture …
-
Deep Learning Based Salient Object Detection for the Detection of Stains and Holes on Patterned Laundry
… the evaluation. Consequently, it is shown that a self-training approach can be used to improve the labels by adding overlooked defects from a model’s predictions, which in turn boosts model performance.
-
Learning Object Detection with Weak Supervision
… in many computer vision applications. However, training deep models typically requires large-scale datasets with elaborate annotations. Collecting and annotating large-scale datasets are laborious, especially for object detection --- a challenging vision task. A promising solution for reducing …
-
Technology Adoption in Secondary Mathematics Teaching in Kenya: An Explanatory Mixed Methods Study
… technology in general (.301), in-service training (.527), and discussions about technology (.161). In the qualitative phase, the participants described how technology training, technology resources, and demographics influenced their decisions to adopt technology in their teaching. Overall, …
Page 1 of 2