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 16 of 16 for “"Inference optimization"”.

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

    … demands, labeled data scarcity, and prohibitive inference costs. This work develops a unified framework of resource-efficient AI techniques and demonstrates their effectiveness across four interconnected research contributions spanning diverse application domains. The first contribution (Chapter …

    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)

  2. Automating inference, learning, and design using probabilistic programming

    … then automating the required computation to draw inferences from the model, for example finding the model parameters likely to give rise to a certain output. By decoupling model specification and inference, PPSs streamline the process of developing and drawing inferences from new models, while …

    oxford Repository record for Automating inference, learning, and design using probabilistic programming (opens in a new tab)

  3. Constrained Statistical Inference for Categorical Data

    … More specifically, we use ordered constrained inference for multinomial data whereby including equality and inequality constraints adds value to our predictions. Using real-world data from the Canadian Community Health Survey (CCHS), the methodology of using constraints showed significant …

    carleton Repository record for Constrained Statistical Inference for Categorical Data (opens in a new tab)

  4. Performance Bottleneck Evaluation of llama.cpp on Jetson and H100

    Efficient large language model inference depends strongly on the interaction between workload shape, numerical precision, serving configuration, and hardware platform. This thesis evaluates the performance bottlenecks of llama.cpp on two contrasting NVIDIA platforms: the datacenter-class H100 GPU …

    chalmers Repository record for Performance Bottleneck Evaluation of llama.cpp on Jetson and H100 (opens in a new tab)

  5. Object Detection Networks at the Edge: Hardware Optimization and Intelligent Transportation Systems Applications

    … in practice. This thesis emphasizes the optimization of DL computation for object detection and proposes designs of real-time Intelligent Transportation Systems (ITS) applications at the edge using hardware accelerators. To optimize DL computation, a novel stride 2 Winograd method is …

    sask Repository record for Object Detection Networks at the Edge: Hardware Optimization and Intelligent Transportation Systems Applications (opens in a new tab)

  6. Automated tuning and analysis of non-reversible parallel tempering

    … a study of how to bridge NRPT and variational inference in order to obtain more effective samplers. To do so, we introduce a generalized annealing path connecting the posterior to an adaptively tuned variational reference, where the reference is tuned to minimize the forward (inclusive) KL …

    ubc Repository record for Automated tuning and analysis of non-reversible parallel tempering (opens in a new tab) Download PDF for Automated tuning and analysis of non-reversible parallel tempering (opens in a new tab)

  7. Boolean models for genetic regulatory networks

    … involved in the genetic regulatory networks: inference, optimization and robustness of the mathe- matical models. Network inference constitutes one of the central goals of genomic signal processing. When inferring rule-based Boolean models of genetic regulations, the same values of predictor …

    tamu Repository record for Boolean models for genetic regulatory networks (opens in a new tab)

  8. Low latency video analytics system with multi-exit neural networks

    … proposed approach aims to have a reduced average inference time as the simple input samples takes an early exit and only the complex samples need more computation offered by all the model layers. The open source CityScapes dataset and the Volvo dataset were used in a number of multi-exit semantic …

    chalmers Repository record for Low latency video analytics system with multi-exit neural networks (opens in a new tab)

  9. ORCHESTRATING MULTI-OBJECTIVE NEURAL NETWORK DEPLOYMENT IN HETEROGENEOUS EDGE SYSTEMS

    … multi-DNN workloads; and (iv) multi-objective optimization methodologies for balancing competing performance metrics in multi-DNN deployment scenarios through three complementary frameworks that address throughput-fairness balancing via reinforcement learning, throughput-power efficiency …

    siu-theses Repository record for ORCHESTRATING MULTI-OBJECTIVE NEURAL NETWORK DEPLOYMENT IN HETEROGENEOUS EDGE SYSTEMS (opens in a new tab)

  10. Applications of Geometry in Optimization and Statistical Estimation

    … models and their influence on statistical inference and asymptotic theory reveal the profound relationship between geometry and statistics. This thesis studies applications of convex and differential geometry to statistical inference, optimization and modelling. We, particularly, …

    waterloo Repository record for Applications of Geometry in Optimization and Statistical Estimation (opens in a new tab)

  11. Klasifikace zvuků pohybujících se objektů pomocí neuronových sítí

    … z hlediska úspěšnosti, výpočetní náročnosti a inference času. Výsledky potvrzují efektivitu zvoleného přístupu i v omezeném výpočetním prostředí.

    brno-tech Repository record for Klasifikace zvuků pohybujících se objektů pomocí neuronových sítí (opens in a new tab)

  12. Υπολογιστική νοημοσύνη - διαχείριση οπτικής πληροφορίας

    … digital image recognition problems and automatic optimization methods on convolutional neural networks, which are now comprising of effective tools for visual information recognition. Regarding the development of smart techniques, the element of locality of visual information was used as well as …

    greece Repository record for Υπολογιστική νοημοσύνη - διαχείριση οπτικής πληροφορίας (opens in a new tab)

  13. Estimation of Treatment Effects in Matching Marketplaces Under Interference

    This thesis develops methodological tools to mitigate interference bias in randomized experiments conducted within matching marketplaces. Interference occurs when treated and untreated units interact, violating standard causal assumptions and introducing bias into treatment effect estimates. The …

    gatech Repository record for Estimation of Treatment Effects in Matching Marketplaces Under Interference (opens in a new tab)

  14. Experimental and Computational Analysis of Mode I Fracture in Wood : Likelihood-based inference and finite element model updating

    … model updating (FEMU) procedure using a Bayesian optimization framework, to establish the optimal set of materialparameters. In addition, two commonly employed cost functions employed in FEMUare reformulated in the context of likelihood-based inference to estimate the uncertainty in the optimal …

    lund Repository record for Experimental and Computational Analysis of Mode I Fracture in Wood : Likelihood-based inference and finite element model updating (opens in a new tab) Download PDF for Experimental and Computational Analysis of Mode I Fracture in Wood : Likelihood-based inference and finite element model updating (opens in a new tab)

  15. RISC-V Low-Power Processors: Enhancing Matrix Computations and Deep Learning Convolution Algorithms

    [ES] En los últimos años, el rápido avance de las tecnologías de aprendizaje profundo (DL) ha transformado significativamente diversos aspectos de nuestra vida. Sin embargo, este progreso conlleva el desafío de procesar enormes volúmenes de datos, lo que exige una potencia computacional muy …

    upv Repository record for RISC-V Low-Power Processors: Enhancing Matrix Computations and Deep Learning Convolution Algorithms (opens in a new tab)