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 13 of 13 for “"Agnostic learning"”.

  1. Sample-Complexity Optimality Under Local Differential Privacy and Related Models

    … explores the sample complexity of estimation and learning tasks under local differential privacy (LDP), an especially strong approach to differential privacy which does not rely on any trusted party to guarantee privacy. We give a new characterization of the sample complexity of answering …

    toronto-retro Repository record for Sample-Complexity Optimality Under Local Differential Privacy and Related Models (opens in a new tab)

  2. Computational foundations of reliable learning

    … lays the computational foundations of reliable learning in binary classification. Classical guarantees assume clean i.i.d. samples from the same distribution on which the classifier is evaluated. Efficient algorithms often assume more: structured marginals, benign noise, no outliers, or stable …

    texas Repository record for Computational foundations of reliable learning (opens in a new tab)

  3. On learning, fairness, and complexity

    In this thesis we study the learning and complexity-theoretic underpinnings of the multigroup fairness framework for prediction algorithms. Multiaccuracy and multicalibration are two primary multigroup fairness notions, which ensure accurate and calibrated predictions, respectively, for every …

    oxford Repository record for On learning, fairness, and complexity (opens in a new tab)

  4. Intractability Results for some Computational Problems

    … show results for some well-studied problems from learning theory and combinatorial optimization. Learning Parities under the Uniform Distribution: We study the learnability of parities in the agnostic learning framework of Haussler and Kearns et al. We show that under the uniform distribution, …

    gatech Repository record for Intractability Results for some Computational Problems (opens in a new tab)

  5. Inherent Limitations of Dimensions for Characterizing Learnability

    The fundamental theorem of statistical learning establishes the equivalence between various notions of both agnostic and realizable Probably Approximately Correct (PAC) learnability for binary classification with the finiteness of the VC dimension. Motivated by these foundational results, this work …

    waterloo Repository record for Inherent Limitations of Dimensions for Characterizing Learnability (opens in a new tab)

  6. AI-infused security: Robust defense by bridging theory and practice

    … with strong theoretical guarantees in the agnostic learning setting. (3) Using AI to Protect Enterprise and Society: We show how AI can be used in real enterprise environment with a novel framework called Virtual Product that predicts potential enterprise cyber threats. Beyond …

    gatech Repository record for AI-infused security: Robust defense by bridging theory and practice (opens in a new tab)

  7. Domain adaptive learning with disentangled features

    … vehicles. However, the success of deep learning methods on visual recognition tasks is highly dependent on access to large-scale labeled datasets, which are expensive and cumbersome to collect. Transfer learning provides a way to alleviate the burden of annotating data, which transfers …

    bu Repository record for Domain adaptive learning with disentangled features (opens in a new tab)

  8. Practical learning in a private and noisy world

    … two fundamental settings in theoretical machine learning where worst-case formulations impose sharp computational hardness barriers: learning under privacy constraints and learning under noise. In both settings, we show that carefully chosen relaxations of the standard models, reflecting …

    bu Repository record for Practical learning in a private and noisy world (opens in a new tab)

  9. Quantum Compression and Quantum Learning via Information Theory

    … of two parts: quantum compression and quantum learning theory. A common theme between these problems is that we study them through the lens of information theory. We first study the task of visible compression of an ensemble of quantum states with entanglement assistance in the one-shot …

    waterloo Repository record for Quantum Compression and Quantum Learning via Information Theory (opens in a new tab)

  10. Towards provably efficient algorithms for learning neural networks

    … Developing provably efficient algorithms for learning commonly used neural network architectures continues to be a core challenge in understanding deep learning. In particular, even the problem of learning very basic architectures remains open. The underlying difficulty arises from the highly …

    texas Repository record for Towards provably efficient algorithms for learning neural networks (opens in a new tab)

  11. An Improved Algorithm for Learning to Perform Exception-Tolerant Abduction

    … of the necessary knowledge by machine learning has been widely found to be highly effective. However, the semantics of learned knowledge are weaker than the usual classical semantics, and this necessitates new formulations of many tasks. We focus on a recently introduced formulation of …

    wustl Repository record for An Improved Algorithm for Learning to Perform Exception-Tolerant Abduction (opens in a new tab)

  12. Learning From Conditional Data Distributions

    Traditional machine learning paradigms often rely on a single global model trained on an entire dataset, aiming for broad generalization across all instances. However, in many real-world applications, the underlying data distribution is heterogeneous, and meaningful predictions often require models …

    wustl Repository record for Learning From Conditional Data Distributions (opens in a new tab)

  13. Exploiting structures of trajectory optimization for efficient optimal motion planning

    … MoE models that outperform the discontinuity-agnostic learning pipeline using standard neural networks. The trajectory learning approach is efficient since it only requires model evaluation to compute a trajectory, reliable since the MoE model is accurate after correctly handling …

    uiuc Repository record for Exploiting structures of trajectory optimization for efficient optimal motion planning (opens in a new tab)