Šį pirmadienį gruodžio 22 d. 13 val. 203 kab. (Akademijos g. 4) DMSTI seminare pranešimą perskaitys DMSTI doktorantė:
Sobia Dagsteer „Unveiling the Black Box: The Role of Explainable AI and Large Language Models in Breast Cancer Diagnosis“
(vadovas prof. dr. Povilas Treigys)
Anotacija: The rising global mortality rate among women due to breast cancer highlights the urgent need for advancements in diagnosis and early detection. Early identification of breast cancer significantly improves patient prognosis and survival outcomes. Artificial Intelligence (AI) has continued to achieve remarkable success in recent years, enhancing the diagnostic and prognostic capabilities of breast cancer detection systems. However, these frameworks often function as "black boxes," making their decision-making process difficult to interpret. Unlike conventional machine learning algorithms, state-of-the-art deep learning models consist of complex interconnected structures, millions of parameters, and a "black box" nature that provides no insights into how they make decisions. These limitations discourage stakeholders from trusting and practically adopting these systems in sensitive domains like healthcare.
To address these challenges, Explainable Artificial Intelligence (XAI) has emerged with the goal of offering explanations that are transparent and interpretable. XAI aims to improve transparency in predictive analysis, which is necessary for the adoption of AI systems in critical domains.
The presentation discusses the ongoing study, explores advanced XAI strategies that balance accuracy with interpretability, including attention-based mechanisms and Large Language Model (LLM) driven explanations. We discuss how LLMs, embedded within XAI systems, act as translational interfaces, and decoding complex model outputs into clinician friendly explanations. We investigate the role of LLMs as an interpretability tool capable of generating meaningful and human understandable explanations of complex model decisions.
[Seminaro antrojo pranešėjo doktoranto dėl nenumatytų techninių problemų, deja, šį kartą nebus].
Maloniai kviečiame dalyvauti! (negalintiems atvykti 'gyvai': https://bit.ly/DMSTI_2025-12-22 )
Tolesni numatomi DMSTI pirmadienio seminarai: 
Tolesni numatomi VU DMSTI pirmadienio seminarai.docx