{"id":{"repo_id":"vilnius","oai_identifier":"oai:vu.lt:elaba:308337840"},"canonical_url":"https://search.ndltd.org/etd/vilnius/oai:vu.lt:elaba:308337840","repository":{"repo_id":"vilnius","name":"Vilnius University","base_url":"https://epublications.vu.lt/oai"},"display":{"title":"Ranka rašytų skaitmenų klasifikavimo uždavinys /","abstract":"This bachelor’s thesis discusses how AI is becoming harder to explain and EU countries are making requirements that in important sectors companies are only allowed to use explainable or interpretable models in high risk situations due to safety and citizen’s right for explanation. Using K-nearest neighbour and deep neural network the study compares both model results in classifying EMNIST “By_Class” dataset in which there are ten outputs for digits and one additional output for non-digit symbols. Along with comparing clean test datasets the study also compares results on tests examples which have specific amount of Gaussian or Salt and Pepper noise applied. The study experiments with improving results with additionally applying noise for training dataset which improves the neural network results in substantial amount. Future research recommends evaluating this dataset’s results using Convoluted neural networks and try to apply explainability methods such as SHAP or LIME on noisy images to see why a neural network got such results.","abstract_html":"This bachelor’s thesis discusses how AI is becoming harder to explain and EU countries are making requirements that in important sectors companies are only allowed to use explainable or interpretable models in high risk situations due to safety and citizen’s right for explanation. Using K-nearest neighbour and deep neural network the study compares both model results in classifying EMNIST “By_Class” dataset in which there are ten outputs for digits and one additional output for non-digit symbols. Along with comparing clean test datasets the study also compares results on tests examples which have specific amount of Gaussian or Salt and Pepper noise applied. The study experiments with improving results with additionally applying noise for training dataset which improves the neural network results in substantial amount. Future research recommends evaluating this dataset’s results using Convoluted neural networks and try to apply explainability methods such as SHAP or LIME on noisy images to see why a neural network got such results.","abstract_has_math":false,"creators":["Povedionok, Robertas,"],"institution":"Institutional Repository of Vilnius University","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":[],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-19T10:12:59Z","subjects":[],"languages":["lit"],"rights":["info:eu-repo/semantics/openAccess"],"rights_urls":[],"identifier_entries":[],"person_identifier_entries":[]},"links":{"outbound_url":"https://repository.vu.lt/VU:ELABAETD308337840&prefLang=en_US","outbound_label":"Repository record","outbound_source":"dc:identifier","record_url":"https://repository.vu.lt/VU:ELABAETD308337840&prefLang=en_US","record_label":"Repository record","record_kind":"record","record_source":"dc:identifier","download_url":null,"download_label":null,"download_source":null,"access_status":"open","embargo_until":null,"access_note":null},"source_record":{"url":"https://epublications.vu.lt/oai?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Avu.lt%3Aelaba%3A308337840","prefix":"oai_dc"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:creator","label":"Author","values":["Povedionok, Robertas,"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2026"]},{"key":"dc:publisher","label":"Institution","values":["Institutional Repository of Vilnius University"]},{"key":"dc:relation","label":"Dc Relation","values":["https://epublications.vu.lt/object/elaba:308337840/308337840.pdf"]},{"key":"dc:type","label":"Dc Type","values":["info:eu-repo/semantics/bachelorThesis"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["lit"]},{"key":"dc:rights","label":"Dc Rights","values":["info:eu-repo/semantics/openAccess"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://repository.vu.lt/VU:ELABAETD308337840&prefLang=en_US"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["This bachelor’s thesis discusses how AI is becoming harder to explain and EU countries are making requirements that in important sectors companies are only allowed to use explainable or interpretable models in high risk situations due to safety and citizen’s right for explanation. Using K-nearest neighbour and deep neural network the study compares both model results in classifying EMNIST “By_Class” dataset in which there are ten outputs for digits and one additional output for non-digit symbols. Along with comparing clean test datasets the study also compares results on tests examples which have specific amount of Gaussian or Salt and Pepper noise applied. The study experiments with improving results with additionally applying noise for training dataset which improves the neural network results in substantial amount. Future research recommends evaluating this dataset’s results using Convoluted neural networks and try to apply explainability methods such as SHAP or LIME on noisy images to see why a neural network got such results."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Ranka rašytų skaitmenų klasifikavimo uždavinys /","Handwritten digits classification task."]}]}],"canonical_facts":{"dc:creator":["Povedionok, Robertas,"],"dc:date":["2026"],"dc:description":["This bachelor’s thesis discusses how AI is becoming harder to explain and EU countries are making requirements that in important sectors companies are only allowed to use explainable or interpretable models in high risk situations due to safety and citizen’s right for explanation. Using K-nearest neighbour and deep neural network the study compares both model results in classifying EMNIST “By_Class” dataset in which there are ten outputs for digits and one additional output for non-digit symbols. Along with comparing clean test datasets the study also compares results on tests examples which have specific amount of Gaussian or Salt and Pepper noise applied. The study experiments with improving results with additionally applying noise for training dataset which improves the neural network results in substantial amount. Future research recommends evaluating this dataset’s results using Convoluted neural networks and try to apply explainability methods such as SHAP or LIME on noisy images to see why a neural network got such results."],"dc:format":["application/pdf"],"dc:identifier":["https://repository.vu.lt/VU:ELABAETD308337840&prefLang=en_US"],"dc:language":["lit"],"dc:publisher":["Institutional Repository of Vilnius University"],"dc:relation":["https://epublications.vu.lt/object/elaba:308337840/308337840.pdf"],"dc:rights":["info:eu-repo/semantics/openAccess"],"dc:title":["Ranka rašytų skaitmenų klasifikavimo uždavinys /","Handwritten digits classification task."],"dc:type":["info:eu-repo/semantics/bachelorThesis"]},"updated_at":"2026-11-19T10:12:59Z"}