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 116 for “"Model Efficiency"”.
-
Probabilistic language models with model efficiency and data efficiency
Probabilistic language models have provided remarkable performance improvements in the domain of natural language processing (NLP). This dissertation presents new approaches addressing three facets of language models from a probabilistic standpoint. Initially, the focus lies on attention-based …
-
Deep Video Understanding with Model Efficiency and Sparse Active Labeling
… we first propose a method to improve network efficiency for video understanding task and then provide methods to improve annotation efficiency for video understanding task. Through these works, we aim to improve the network efficiency as well as data annotation efficiency, as an effort to …
-
A comparison of techniques used in rainfall-runoff models : model efficiency
A suite of three underlying rainfall-runoff modeling techniques is applied to two data sets and the results used to compare model efficiencies for selected events. Linear regression, unit hydrograph, and quasi-physically based models make up the modeling suite. The two data sets come from a 7.2 KM …
-
Enhancing generative model efficiency and control for data generation, reinforcement learning, and robotics
The rapid advancement of generative models has driven their application across a wide range of domains, from image synthesis to complex decision-making systems. This dissertation focuses on enhancing both the efficiency and controllability of generative models, addressing significant challenges …
-
Hierarchical and Hardware-Aware Optimization for Enhancing AI Model Efficiency: From Bits to Modules to Models
… the remarkable advancements of AI foundation models, such as large language models (LLMs), in numerous tasks and applications, deploying these powerful models on everyday devices remains challenging due to their growing computational and memory demands. This challenge hinders the realization …
-
Efficient Deep Learning for Image Applications
… developers, and machine learning end users, is inefficiency of DL methods: intolerable training and inference time, exhausted computing resources, and unsustainable power consumption.</p><p>To tackle the inefficiency issues, tons of DL efficiency methods have been proposed to improve efficiency …
-
EFFICIENCY IN THE ERA OF FOUNDATION MODEL
The era of foundation models has created unprecedented efficiency challenges across two critical dimensions: model efficiency and data efficiency. This thesis provides a comprehensive investigation of both dimensions, demonstrating that their combination offers the most practical path to deploying …
-
Development and Evaluation of a Gis-Based Spatially Distributed Unit Hydrograph Model
… a spatially distributed unit hydrograph (SDUH) model that directly reflects spatial variation in the watershed in generating runoff hydrographs. The SDUH model is a time-area unit hydrograph technique that uses a geographic information system (GIS) to develop a cumulative travel time map of the …
-
On the Resource Efficiency of Language Models
Large language models (LLMs) have revolutionized a wide range of natural language processing tasks. However, their practical utility faces resource challenges in two dimensions: data efficiency and model efficiency. For post-training, LLMs face data curation challenges where high-quality labeled …
-
Robust and Efficient Learning for Visual Perception
… for both tasks, designed to rigorously evaluate model efficiency and stability in real-world environments. Experimental results demonstrate that these methods significantly improve computational throughput and parameter efficiency while maintaining state-of-the-art accuracy, providing scalable …
-
Efficient LLM training and inference with contextual sparsity
Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2027-05-01
-
Toward Efficient Deep Learning for Computer Vision Applications
… to require larger datasets and more complex models, leading to heightened resource consumption across all fronts. Regrettably, meeting these requirements proves challenging in many real-life scenarios. First, both data collection and labeling processes entail substantial labor and time …
-
TOWARDS EFFICIENT AND GENERALIZABLE 3D HUMAN POSE ESTIMATION AND ITS APPLICATIONS
… motions and large pose deformation. To improve model efficiency, we propose a novel point-based representation for monocular multi-person body mesh estimation, which achieves real-time inference speed. Furthermore, we present an inference stage optimization pipeline, which helps improve model …
-
Digitální stín tepelné soustavy
Tato diplomová práce se zabývá návrhem modelu výměníkové stanice, identifikací jeho parametrů a jeho nasazením do provozu ve formě digitálního stínu. Teoretická část popisuje metody a možnosti modelování průmyslových systémů a fyzický model výměníkové stanice. Praktická část obsahuje popis řízení, …
-
Pruning visual transformers to increase model compression and decrease inference time
… reductions in inference time without sacrificing model accuracy. Different pruning rates were evaluated, demonstrating a trade-off between training speed and model compression. Slower pruning rates allowed for better convergence to higher accuracy levels and more efficient model recovery. …
-
Predicting suspended sediment in Illinois rivers using dimensionless rating curves
… of the hydrograph yielded predictive sediment models. The data from all watersheds were then combined and normalized to create a dimensionless sediment rating curve for Illinois. Model efficiency was low, but improved when data was subdivided by landuse, channel slope, or drainage area. …
-
Sensitivity analysis across scales and watershed discretization schemes using ARDBSN hydrological model and GIS
The goal of this study is to assess ARDBSN model performance across watershed sizes, and to assess the effect of using different mapset resolutions and basin configurations on runoff volume. Arid basin (ARDBSN) (Lane, 1982) is a distributed parameter, continuous hydrologic simulation model designed …
-
Architecture and allocation considerations for group expert systems
… group expert system on a Local Area Network. A model system called GESP (Group Expert System Prototype) was implemented in Prolog on a microcomputer LAN to be used as a working platform. From observations of the model, conclusions have been drawn concerning: (1) the architecture of the expert …
-
Algorithm-Hardware Co-Design Towards Efficient and Robust Edge Vision Applications
… one of the two aforementioned challenges (i.e., efficiency or adversarial robustness) while neglecting or even sacrificing the other. To this end, we propose a 2-in-1 Accelerator, an integrated algorithm-hardware co-design framework aiming at winning both the adversarial robustness and efficiency …
-
TOWARDS EFFICIENT OBJECT DETECTION WITH DEEP LEARNING
… detection. Concretely, we take two perspectives, model efficiency and data efficiency. We propose detection knowledge distillation and feature compression technique to accelerate model inference. We establish few-shot adaptation detection and long-tail detection frameworks to achieve efficient …
Page 1 of 6