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 11 of 11 for “"Masked Autoencoding"”.

  1. Self-Supervised Masked Autoencoding Meets Federated Learning for Electric Vehicle Battery State-of-Health Estimation

    EVs live and die by their batteries. To keep drivers safe and confident in their vehicles, we need efficient, accurate, and private ways to track each battery's SoH. But, EV labelled data is scarce, sharing raw data raises privacy flags, and big models strain on-board hardware. This thesis tackles …

    mcmaster Repository record for Self-Supervised Masked Autoencoding Meets Federated Learning for Electric Vehicle Battery State-of-Health Estimation (opens in a new tab)

  2. Self-supervised Pretraining of Vision Transformers for Earth Observation

    … (ViTs) using self-supervised learning. Using masked autoencoding, I develop one of the first ViTs pretrained on RS data—called SatViT. Next, I develop an improved foundation model—called SatViT-V2—and evaluate it on out-of-distribution data. I show that these models are more robust to …

    carleton Repository record for Self-supervised Pretraining of Vision Transformers for Earth Observation (opens in a new tab)

  3. Advanced Transformer-based Models for Medical Image Analysis

    … a self-supervised foundation model based on masked autoencoding is pretrained on large-scale unlabelled data. The model demonstrates improved robustness and data efficiency across multiple downstream diagnostic tasks.

    carleton Repository record for Advanced Transformer-based Models for Medical Image Analysis (opens in a new tab)

  4. Advancing Differential Equation-based Modeling in Computational Medicine and Deep Knowledge Transfer for Multi-Modal GeoAI

    … a self-supervised learning framework integrates masked autoencoding and soft-alignment pseudo-labeling to fine-tune foundational models such as a recent 2023 NASA/IBM geospatial foundation model. This approach improves segmentation accuracy and adaptability across RS modalities, including RGB, …

    houston Repository record for Advancing Differential Equation-based Modeling in Computational Medicine and Deep Knowledge Transfer for Multi-Modal GeoAI (opens in a new tab) Embargoed until 2027-05-01

  5. Robust and Trustworthy Medical Time Series Analysis: From Generative Denoising to Interpretable Decision Support

    … joint-embedding contrastive learning and masked autoencoding to learn robust, occlusion-invariant representations from unlabeled medical time series. This method captures both local events and global temporal dependencies, enabling accurate classification across multiple clinical tasks …

    arizona-thes Repository record for Robust and Trustworthy Medical Time Series Analysis: From Generative Denoising to Interpretable Decision Support (opens in a new tab)

  6. Toward the Generalization of Vision Models

    … tasks. Additionally, I introduce efficient masked autoencoding techniques for enhanced representation learning. Finally, I propose Sequential Modeling Enables Scalable Learning for Large Vision Models, a methodology designed to effectively leverage large volumes of real-world visual data, …

    jhu Repository record for Toward the Generalization of Vision Models (opens in a new tab)

  7. Towards Accurate and Fair Deepfake Detection

    … transformers pre-trained by a self-supervised masked autoencoding setup. Our method consists of two distinct components, one of which focuses on learning spatial information from individual RGB frames of the video, while the other learns temporal consistency information from optical flow fields …

    queens Repository record for Towards Accurate and Fair Deepfake Detection (opens in a new tab)

  8. Toward learning representations for movements

    … a hybrid self-supervised framework that combines masked autoencoding (MAE) with joint embedding prediction architectures (JEPA). While JEPA have demonstrated strong performance on dense video data, they tend to underperform on sparse skeletal data. By integrating coordinate reconstruction and …

    bu Repository record for Toward learning representations for movements (opens in a new tab)

  9. Resource-efficient AI for real-world applications: Model compression, sample-efficient learning, and inference optimization across traffic safety, quantum systems, and sequential decision making

    … learning for quantum systems through Hamiltonian-Masked Autoencoding (HMAE), reducing labeled data requirements by 3-5× and achieving 85.3% accuracy in phase classification with only 10 labeled examples—addressing the critical challenge of expensive quantum simulations. The third contribution …

    iastate Repository record for Resource-efficient AI for real-world applications: Model compression, sample-efficient learning, and inference optimization across traffic safety, quantum systems, and sequential decision making (opens in a new tab)

  10. Multi-modal Data-efficient Learning for 3D Machine Vision

    … To align features across modalities, a two-stage masked token prediction framework is proposed. It incorporates semantic masks from segmentation models and uses group-balanced reweighting to mitigate long-tailed class distributions, facilitating effective cross-modal feature alignment. Finally, …

    clemson Repository record for Multi-modal Data-efficient Learning for 3D Machine Vision (opens in a new tab)

  11. Multimodal models for understanding and assisting skilled activities

    … frames and binaural audio. The method trains a masked autoencoding framework to reconstruct missing segments of binaural audio from the visual stream and the remaining audio, together with a spatial audio masking strategy that encourages the model to learn robust audio-visual spatial …

    texas Repository record for Multimodal models for understanding and assisting skilled activities (opens in a new tab)