Seyed Mohammad Hossein Hashemi

Seyed Mohammad Hossein Hashemi, Leila Safari, Mohsen Hooshmand et al.

Reliable diagnosis of brain tumors remains challenging due to low clinical incidence rates of such cases. However, this low rate is neglected in most of proposed methods. We propose a clinically inspired framework for anomaly-resilient tumor detection and classification. Detection leverages YOLOv8n fine-tuned on a realistically imbalanced dataset (1:9 tumor-to-normal ratio; 30,000 MRI slices from 81 patients). In addition, we propose a novel Patient-to-Patient (PTP) metric that evaluates diagnostic reliability at the patient level. Classification employs knowledge distillation: a Data Efficient Image Transformer (DeiT) student model is distilled from a ResNet152 teacher. The distilled ViT achieves an F1-score of 0.92 within 20 epochs, matching near teacher performance (F1=0.97) with significantly reduced computational resources. This end-to-end framework demonstrates high robustness in clinically representative anomaly-distributed data, offering a viable tool that adheres to realistic situations in clinics.

Mehrnoush Shamsfard, Zahra Saaberi, Mostafa Karimi manesh et al.

Research on evaluating and analyzing large language models (LLMs) has been extensive for resource-rich languages such as English, yet their performance in languages such as Persian has received considerably less attention. This paper introduces FarsEval-PKBETS benchmark, a subset of FarsEval project for evaluating large language models in Persian. This benchmark consists of 4000 questions and answers in various formats, including multiple choice, short answer and descriptive responses. It covers a wide range of domains and tasks,including medicine, law, religion, Persian language, encyclopedic knowledge, human preferences, social knowledge, ethics and bias, text generation, and respecting others' rights. This bechmark incorporates linguistics, cultural, and local considerations relevant to the Persian language and Iran. To ensure the questions are challenging for current LLMs, three models -- Llama3-70B, PersianMind, and Dorna -- were evaluated using this benchmark. Their average accuracy was below 50%, meaning they provided fully correct answers to fewer than half of the questions. These results indicate that current language models are still far from being able to solve this benchmark