{"id":"https://openalex.org/W7134164008","doi":"https://doi.org/10.1109/bigdata66926.2025.11402394","title":"Vision Token Reduction via Attention-Driven Self-Compression for Efficient Multimodal Large Language Models","display_name":"Vision Token Reduction via Attention-Driven Self-Compression for Efficient Multimodal Large Language Models","publication_year":2025,"publication_date":"2025-12-08","ids":{"openalex":"https://openalex.org/W7134164008","doi":"https://doi.org/10.1109/bigdata66926.2025.11402394"},"language":null,"primary_location":{"id":"doi:10.1109/bigdata66926.2025.11402394","is_oa":false,"landing_page_url":"https://doi.org/10.1109/bigdata66926.2025.11402394","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Big Data (BigData)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Omer Faruk Deniz","orcid":null},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Omer Faruk Deniz","raw_affiliation_strings":["The University of Texas at Dallas,Richardson,TX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas,Richardson,TX","institution_ids":["https://openalex.org/I162577319"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Ruiyu Mao","orcid":null},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ruiyu Mao","raw_affiliation_strings":["The University of Texas at Dallas,Richardson,TX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas,Richardson,TX","institution_ids":["https://openalex.org/I162577319"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Ruochen Li","orcid":null},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ruochen Li","raw_affiliation_strings":["The University of Texas at Dallas,Richardson,TX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas,Richardson,TX","institution_ids":["https://openalex.org/I162577319"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Yapeng Tian","orcid":null},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yapeng Tian","raw_affiliation_strings":["The University of Texas at Dallas,Richardson,TX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas,Richardson,TX","institution_ids":["https://openalex.org/I162577319"]}]},{"author_position":"last","author":{"id":null,"display_name":"Latifur Khan","orcid":null},"institutions":[{"id":"https://openalex.org/I162577319","display_name":"The University of Texas at Dallas","ror":"https://ror.org/049emcs32","country_code":"US","type":"education","lineage":["https://openalex.org/I162577319"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Latifur Khan","raw_affiliation_strings":["The University of Texas at Dallas,Richardson,TX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Dallas,Richardson,TX","institution_ids":["https://openalex.org/I162577319"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I162577319"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4544","last_page":"4551"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.6331999897956848,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.6331999897956848,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.13429999351501465,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.04410000145435333,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.5059999823570251},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.33629998564720154},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.3109999895095825},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.3089999854564667},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.2847000062465668}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6725000143051147},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5315999984741211},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.5059999823570251},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.36629998683929443},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.33629998564720154},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3330000042915344},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.3109999895095825},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.3089999854564667},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2847000062465668},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.27639999985694885},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.2709999978542328},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.25220000743865967}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/bigdata66926.2025.11402394","is_oa":false,"landing_page_url":"https://doi.org/10.1109/bigdata66926.2025.11402394","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE International Conference on Big Data (BigData)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3740812561","display_name":"PURPOSE: ADVANCEMENTS IN AI HAVE BROUGHT CHALLENGES THAT TOUCH ON THE RELIABILITY, SECURITY AND PRIVACY OF AI MODELS. AS AI IS ALREADY RELIED UPON IN HEALTHCARE SCENARIOS FOR DECISION SUPPORT THESE CONCERNS ARE HIGHLIGHTED EVEN MORE. THIS PROJECT ENDEAVORS TO EXPLORE AND TEST POSSIBLE SOLUTIONS TO ADDRESS SOME OF THESE CHALLENGES, RANGING FROM METHODS TO PROVIDE INSIGHT INTO OTHERWISE OPAQUE AI SYSTEM COMPONENTS; TESTING METHODS TO PREVENT MALICIOUS ACTORS BYPASSING THE GUARDRAILS FOR MALICIOUS PURPOSES; TO IDENTIFYING A WAY TO ADDRESS A KEY CHALLENGE AROUND THE EXPLAINABILITY OF AI DECISIONS TO THE AI USERS; AND METHODS FOR CORRECTING AI SYSTEM ERRORS.  ACTIVITIES TO BE PERFORMED:  THE RESEARCH TEAM WILL MODEL AGNOSTIC BLACK-BOX DEFENSE FOR SECURING LARGE LANGUAGE MODELS (LLM) BY TESTING A TECHNIQUE TO EVALUATE PROMPTS AGAINST KNOWN HARMFUL PROMPTS, KNOWN SAFE PROMPTS, AS WELL AS A METHOD TO DETECT AND EVALUATE WHETHER THERE IS A LIKELIHOOD THAT THE SEQUENTIAL PROMPTS ARE BEING USED TO BYPASS ANY GUARDRAILS. THEY PLAN TO INVESTIGATE WAYS TO DETECT AND INTERPRET ERRORS (HALLUCINATIONS) IN LLMS INCLUDING UNDERSTANDING WHERE THE ERROR OCCURRED AND THE CAUSES OF THESE ERRORS. THEY WILL EXPLORE USING TRUSTED EXECUTION ENVIRONMENTS TO MINIMIZE DATA IN PURSUIT OF PRIVACY-PRESERVING, FAIR AND EXPLAINABLE LLM BASED CHATBOTS FOR HEALTHCARE. TOWARDS EXPLAINABILITY THEY WILL INVESTIGATE TOOLS THAT IDENTIFY THE ATTRIBUTES THAT WHERE THE MOST IMPACTFUL IN THE DECISION-MAKING PROCESS. THEY WILL INVESTIGATE WHETHER AI CAN BE USED IN A DEFENSIVE USE CASE WHERE THE AI IS USED TO RECOGNIZE ATTACKS USING THE MITRE ATT&CK FRAMEWORK, AND THEN PERFORM DECEPTION TECHNIQUES IN ORDER TO MONITOR THE ATTACKER THAT HAS BEEN DECEIVED INTO FOCUSING ON A NON-IMPORTANT TARGET OR FEEDING THE ATTACKER MISINFORMATION. THEY WILL ALSO INVESTIGATE THE USE OF LLMS TO IDENTIFY THE POSSIBLE PLAYS THAT CAN BE USED FOR RESPONSES TO A DETECTED ATTACK IN DIFFERENT SCENARIOS. FINALLY, THEY WILL USE THE RESULTS FROM ALL THE ABOVE RESEARCH TO CREATE AN AI GOVERNANCE FRAMEWORK FOR LLM BASED CHATBOTS. EXPECTED OUTCOMES: THIS RESEARCH HAS THE POTENTIAL FOR THE HEALTHCARE INDUSTRY ONCE PROVEN TO ALLOW PATIENTS TO USE CHATBOTS SAFELY AND PREVENT DANGEROUS SITUATIONS SUCH AS TRYING TO DO SELF-HARM TO ONESELF. BEING ABLE TO EXPLAIN HOW DECISIONS WERE MADE OR CONCLUSIONS REACHED IN AN UNDERSTANDABLE WAY WILL ENABLE TRUST IN THOSE RESULTS BY DOCTORS WHICH WILL ALLOW BROADER ADOPTION OF THESE AI TECHNOLOGIES. THE RESEARCH FOCUSED ON FAIRNESS AND PRIVACY ARE ALSO KEY TO ENABLING TRUST IN AI TECHNOLOGIES. FINALLY, THE RESEARCH IN THE USE OF AI FOR CYBERSECURITY MAY OFFER KEY INSIGHTS INTO POTENTIAL APPROACHES FOR RESPONDING TO THE EVER-GROWING ATTACKS ON INFORMATION SYSTEMS AND TECHNOLOGIES. THE METHODOLOGY RESULTING FROM THIS PROJECT WILL BE SHARED FREELY WITH THE BROAD COMMUNITY OF RESEARCHERS, EDUCATORS, AND ANALYSTS VIA PUBLICLY AVAILABLE PYTHON PROJECTS ON GITHUB. ALSO, THEY ANTICIPATE THAT THEY WILL OUTPUT ROBUST ML TOOLS DEVELOPED IN PYTHON AND SHARE THESE PYTHON TOOLS OPENLY WHICH WILL BENEFIT A BROAD COMMUNITY. INTENDED BENEFICIARIES: BUSINESSES AND ORGANIZATIONS TRYING TO LEVERAGE AI-TOOLS IN A SECURE AND PRIVACY-PRESERVING MANNER. PROJECT DELIVERABLES, TOOLS AND GUIDELINES WILL BE AVAILABLE FOR ORGANIZATIONS TO USE. THE PROPOSED CENTER FOR SECURE AND TRUSTWORTHY AI WILL PROVIDE SUPPORT FOR AI-ORIENTED START-UPS IN THE DALLAS-FORT WORTH METROPOLITAN AREA.  SUBRECIPIENT ACTIVITIES: THERE ARE NO PLANNED SUBAWARDS.","funder_award_id":"60NANB24D143","funder_id":"https://openalex.org/F4320332178","funder_display_name":"National Institute of Standards and Technology"},{"id":"https://openalex.org/G3830173093","display_name":null,"funder_award_id":"2138259,2138286,2138307,2137603,2138296","funder_id":"https://openalex.org/F4320306076","funder_display_name":"National Science Foundation"},{"id":"https://openalex.org/G8031005220","display_name":null,"funder_award_id":"69A3552344812","funder_id":"https://openalex.org/F4320309106","funder_display_name":"Clemson University"},{"id":"https://openalex.org/G8939998072","display_name":null,"funder_award_id":"CIS240609","funder_id":"https://openalex.org/F4320312143","funder_display_name":"National Centre for Supercomputing Applications"}],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320309106","display_name":"Clemson University","ror":"https://ror.org/037s24f05"},{"id":"https://openalex.org/F4320312143","display_name":"National Centre for Supercomputing Applications","ror":"https://ror.org/03r10zj06"},{"id":"https://openalex.org/F4320332178","display_name":"National Institute of Standards and Technology","ror":"https://ror.org/05xpvk416"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":9,"referenced_works":["https://openalex.org/W4226407477","https://openalex.org/W4312910119","https://openalex.org/W4402727764","https://openalex.org/W4403793633","https://openalex.org/W4403939369","https://openalex.org/W4409346467","https://openalex.org/W7133193597","https://openalex.org/W7133196460","https://openalex.org/W7133227460"],"related_works":[],"abstract_inverted_index":null,"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2026-02-18T00:00:00"}