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 18 of 18 for “"Data distillation"”.
-
Cyber-physical data distillation in a sensor rich world
… will be dominated by vast amount of sensing data. This thesis attacks a grand challenge in this vision: how to extract human-consumable information from such data? The prime contribution of this work is a framework called FusionSuite that facilitates the development of future applications and …
-
Knowledge Distillation as a Solution for Domain Transfer
… in fact verification tasks. We present a data distillation technique for delexicalization, which we then combine with a model distillation method to prevent aggressive data distillation. We show that by using our solution, not only does the performance of an ex- isting state-of-the-art …
-
On the Effect of Data on Image Classification Tasks
… various domains, but its reliance on large-scale datasets and complex models raises concerns about sustainability and efficiency. This thesis aims to leverage data potential in image classification tasks, focusing on two critical aspects: efficient learning and multi-label representation learning. …
-
Solvent recovery from a complex process solvent stream
… process utilizing vapor-liquid equilibrium data, distillation and extraction procedures. Considerable effort was expended to develop precise and reliable assay methods. A process which met the necessary requirements of maximum solvent recovery with minimum waste disposal, was developed.</p>
-
Learning to Learn: Meta-Knowledge Distillation
… performance predictions, and 3) a novel data distillation method using Sparse Gaussian Processes and a meta-data selection algorithm for task-specific optimization. Our first contribution provides a formal definition of transfer learning, which leverages knowledge gained from previous …
-
Multimodal Learning for 3D Perception in Autonomous Driving Environments
… strengths, with lidar providing accurate depth data but returning sparse point clouds, while cameras deliver dense, color-rich images but lack depth information. In addition, unlike the large-scale datasets available for images across multiple domains, lidar datasets are less diverse, which …
-
Towards a Trustworthy Cyberspace: A Multimethod Approach
… and beyond confirmation bias. Essay II develops data distillation methods that improve neural networks’ performance in out-of-domain settings to improve fake news detection. Essay III studies the impact of nonverbal tokens on the spread of information. This dissertation offers several …
-
Differentially private neural tangent kernels for privacy-preserving data generation and distillation
With the increasing interest in Deep Learning, data safety issues have become more prevalent as we rely more on Artificial Intelligence. Adversaries can easily obtain sensitive information through various attacks, this dramatically discourages patients and clients from contributing invaluable data …
-
Grounding Perception and Reasoning in Multimodal Models
… continued improvements from scaling model and data size. However, this reliance on web data has not yet translated into fully reliable multimodal systems; these models remain prone to hallucinating content and failing to correctly ground their reasoning in images and videos. This disconnect …
-
Towards Trustworthy Deep Graph Learning
… tools for modeling complex graph-structured data across fields such as bioinformatics, finance, and social network analysis. However, despite their success, DGL faces critical challenges related to trustworthiness--a quality encompassing robustness against adversarial attacks, explainability …
-
Task Distillation: Transforming Reinforcement Learning into Supervised Learning
Recent work in dataset distillation focuses on distilling supervised classification datasets into smaller, synthetic supervised datasets in order to reduce per-model costs of training, to provide interpretability, and to anonymize data. Distillation and its benefits can be extended to a wider array …
-
Iterative Data Augmentation For Enhancing Deep Learning Performance With Limited Training Data
… heavily depends on large, well annotated datasets. In practice, data are often limited in size, leading to overfitting, poor generalization, and degraded model robustness and performance. Moreover, conventional augmentation techniques are typically static in nature, lack adaptability …
-
Data Sharing and Retrieval of Manufacturing Processes
… pool their resources to acquire large amounts of data that can then be used in machine learning tasks. Despite the potential to speed up training and deployment and improve decision-making through data-sharing, rising privacy concerns are slowing the spread of such technologies. As businesses are …
-
Efficient data to decision pipelines for embedded and social sensing
… presents results of our studies in making data to decision pipelines for embedded and social sensing efficient. Due to the pervasive presence of wired sensors, wireless sensors, and mobile devices, the amount of data about the physical world (environmental measurements, traffic, etc.) and …
-
Privacy-preserving synthetic image data generation and classification
… reliability of medical findings. Sensitive image data and deep learning’s high memorization capacity pose privacy risks, particularly for medical images containing sensitive private information. De-anonymization does not work due to the re-identification risk and reduced utility. So, we developed …
-
Clean Transportation: Effects of Heterogeneity and Technological Progress on EV Costs and CO2 Abatement, and Assessment of Public EV Charging Stations
… the public EV charging stations using empirical data collected in the City of Rochester. A data distillation is first conducted for the raw data to construct the daily charging profiles of the EV users. A pattern analysis is then performed to identify 5 distinct and homogenous clusters of daily …
-
On quantifying the quality of information in social sensing
… humans play a critical role in the sensing or data collection process. Social sensing has emerged as a new paradigm for sensory data collection, which is motivated by the proliferation of mobile platforms equipped with a variety of sensors (e.g., GPS, camera, microphone, motion and etc.) in the …