5 papers
ChainDoRA: Tensor-Train Factorized Weight-Decomposed Low-Rank Adaptation for Parameter-Efficient LLM Fine-Tuning
Ashfak Yeafi, Mehedi Hasan, Md Khairul Islam
Parameter-efficient fine-tuning (PEFT) adapts large language models (LLMs) to downstream tasks while updating only a small fraction of their pretrained parameters. Low-Rank Adaptat…
TAP-Path: Task-Adaptive Structural and Token Pruning for Efficient and Trustworthy Pathology Foundation Models
Mehedi Hasan, Ashfak Yeafi, Md Khairul Islam
Pathology foundation models improve transferable representation learning for histopathology, but recent gains often rely on encoders with hundreds of millions of parameters and hig…
Linear Multi-Timescale Retention as a Memory-Efficient Vision-Language Bridge
Ashfak Yeafi, Mehedi Hasan, Md Khairul Islam
Vision-Language Models (VLMs) face a critical computational bottleneck when processing high-resolution imagery due to the memory complexity of Softmax Multi-Head Attention…
PrivEraserVerify: Efficient, Private, and Verifiable Federated Unlearning
Parthaw Goswami, Md Khairul Islam, Ashfak Yeafi
Federated learning (FL) enables collaborative model training without sharing raw data, offering a promising path toward privacy preserving artificial intelligence. However, FL mode…
SwinTextUNet: Integrating CLIP-Based Text Guidance into Swin Transformer U-Nets for Medical Image Segmentation
Ashfak Yeafi, Parthaw Goswami, Md Khairul Islam +1
Precise medical image segmentation is fundamental for enabling computer aided diagnosis and effective treatment planning. Traditional models that rely solely on visual features oft…