--- OA_place: publisher _id: '21854' abstract: - lang: eng text: "As neural-network-based models grow both in size and popularity, interest has grown in making the models smaller and more efficient to train. To that end, many methods have been proposed to prune models by reducing their number of nonzero parameters. Additionally, parameter-efficient fine-tuning, in which a much smaller number of parameters than the total contained in the model is updated during training, has become very popular, especially in the space of Large Language Models. At the same time, the increasingly routine deployment of machine learning in real-world applications has spurred a drive to make them more trustworthy - in the sense of, among other things, being unbiased, interpretable, and editable. In this thesis, we examine the interplay between efficiency and trustworthiness.\r\n\r\nFirst, we analyze the effects of model pruning on bias in computer vision models, demonstrating that increased sparsity leads to greater bias, largely as a function of increased model uncertainty in marginal cases. Based on this observation, we propose several bias mitigation techniques. Then, we demonstrate that example-specific model pruning can improve model interpretation methods while improving pruning efficiency to make example-specific model pruning feasible in real time. Then, we investigate the effectiveness of parameter-efficient and data-efficient model personalization via fine-tuning, demonstrating that it is highly feasible with very small computational and data resources. Finally, we consider efficiency in editing model knowledge using a custom synthetic data framework, demonstrating that parameter-efficient, low-rank fine-tuning frequently outperforms full-rank fine-tuning, and, additionally, that restricting which model blocks are fine-tuned frequently improves results. Together, the results in this thesis provide new insights and techniques for combining trustworthiness and efficiency during neural network inference and training.\r\n\r\n-----------------“In reference to IEEE copyrighted material which is used with permission in this thesis, the IEEE does not endorse any of [name of university or educational entity]’s products or services. Internal or personal use of this material is permitted. If interested in reprinting/republishing IEEE copyrighted material for advertising or promotional purposes or for creating new collective works for resale or redistribution, please go to http://www.ieee.org/publications_standards/publications/rights/rights_link.html to learn how to obtain a License from RightsLink. If applicable, University Microfilms and/or ProQuest Library, or the Archives of Canada may supply single copies of the dissertation.”" acknowledged_ssus: - _id: ScienComp acknowledgement: "The research in this Ph.D. was funded in whole\r\nor in part by the Austrian Science Fund (FWF) W1260-N35 (Vienna Graduate School for\r\nComputational Optimization). For open access purposes the author has applied a CC BY\r\npublic copyright license to any author accepted manuscript version arising from this submission\r\nwherever possible. Additionally, I am grateful to Alois Schlögl, Waleed Khalid, and the rest of\r\nthe ISTA Scientific Computing team for building and maintaining the infrastructure I used\r\nto run experiments. I’m also deeply grateful to the Alistarh group’s administrative assistant,\r\nChristine Francois, who always deals with our nonsense with common sense and a smile.\r\n" alternative_title: - ISTA Thesis article_processing_charge: No author: - first_name: Eugenia B full_name: Iofinova, Eugenia B id: f9a17499-f6e0-11ea-865d-fdf9a3f77117 last_name: Iofinova orcid: 0000-0002-7778-3221 citation: ama: Iofinova EB. On the utility and effects of efficiency in artificial neural networks. 2026. doi:10.15479/AT-ISTA-21854 apa: Iofinova, E. B. (2026). On the utility and effects of efficiency in artificial neural networks. Institute of Science and Technology Austria. https://doi.org/10.15479/AT-ISTA-21854 chicago: Iofinova, Eugenia B. “On the Utility and Effects of Efficiency in Artificial Neural Networks.” Institute of Science and Technology Austria, 2026. https://doi.org/10.15479/AT-ISTA-21854. ieee: E. B. Iofinova, “On the utility and effects of efficiency in artificial neural networks,” Institute of Science and Technology Austria, 2026. ista: Iofinova EB. 2026. On the utility and effects of efficiency in artificial neural networks. Institute of Science and Technology Austria. mla: Iofinova, Eugenia B. On the Utility and Effects of Efficiency in Artificial Neural Networks. Institute of Science and Technology Austria, 2026, doi:10.15479/AT-ISTA-21854. short: E.B. Iofinova, On the Utility and Effects of Efficiency in Artificial Neural Networks, Institute of Science and Technology Austria, 2026. corr_author: '1' date_created: 2026-05-11T08:43:22Z date_published: 2026-05-11T00:00:00Z date_updated: 2026-05-19T11:20:28Z day: '11' ddc: - '000' degree_awarded: PhD department: - _id: GradSch - _id: DaAl doi: 10.15479/AT-ISTA-21854 file: - access_level: closed checksum: 2e148dad920e3f9b7c32796e0ba2e5f7 content_type: application/zip creator: eiofinov date_created: 2026-05-11T08:36:01Z date_updated: 2026-05-11T08:36:01Z file_id: '21856' file_name: EIofinova_thesis_FinalVersion.zip file_size: 28479571 relation: source_file - access_level: open_access checksum: b10c2933f386f532b2dbf28b19c5525c content_type: application/pdf creator: eiofinov date_created: 2026-05-13T13:10:48Z date_updated: 2026-05-13T13:10:48Z file_id: '21877' file_name: 2026_Iofinova_Eugenia_Thesis.pdf file_size: 18137757 relation: main_file success: 1 file_date_updated: 2026-05-13T13:10:48Z has_accepted_license: '1' language: - iso: eng month: '05' oa: 1 oa_version: Published Version page: '237' project: - _id: 9B9290DE-BA93-11EA-9121-9846C619BF3A grant_number: W1260-N35 name: Vienna Graduate School on Computational Optimization publication_identifier: issn: - 2663-337X publication_status: published publisher: Institute of Science and Technology Austria related_material: record: - id: '14771' relation: part_of_dissertation status: public - id: '18121' relation: part_of_dissertation status: public - id: '21858' relation: part_of_dissertation status: public - id: '21859' relation: part_of_dissertation status: public - id: '21857' relation: part_of_dissertation status: public status: public supervisor: - first_name: Dan-Adrian full_name: Alistarh, Dan-Adrian id: 4A899BFC-F248-11E8-B48F-1D18A9856A87 last_name: Alistarh orcid: 0000-0003-3650-940X title: On the utility and effects of efficiency in artificial neural networks type: dissertation user_id: 8b945eb4-e2f2-11eb-945a-df72226e66a9 year: '2026' ...