arXiv subjects

Mehmet Emre Akbulut

Publications and source records attributed to Mehmet Emre Akbulut.

3 recordsLinked to original sources

arXiv2026

MemFLoRA: Memory-Floor LoRA for CNN Adaptation at the Edge

On-device learning is necessary when the model encounters user-,sensor-, or environment-specific shifts after deployment. Although parameter-efficient fine-tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA) variants, enable efficient adaptation at the edge, the limiting resource for Convolutional Neural Network (CNN) adaptation is often not the number of trainable parameters but the activation state that must be retained until the backward pass. This paper introduces Memory-Floor LoRA (MemFLoRA), a low-rank CNN adapter built around a memory-first design principle rather than a direct application of transformer-oriented LoRA. Instead of merely reducing trainable weights, we define an activation-memory-floor criterion: trainable backward computations must not depend on full-width layer inputs. The resulting adapter freezes the down-projection, trains a scale-matched up-projection, and combines eval-mode backbone normalization with activation-minimal backward rules, reducing saved state to the low-rank branch. Evaluated on three Human Activity Recognition (HAR) datasets and two CNN backbones under subject, body-location, and sensor-placement shifts, MemFLoRA reduces saved-activation memory by 98.5-98.7% and peak training-state memory by 94.9-97.3% relative to full fine-tuning, while matching or exceeding CNN PEFT baselines.

cs.LG↗

arXiv2025

InfoQ: Mixed-Precision Quantization via Global Information Flow

Mixed-precision quantization (MPQ) is crucial for deploying deep neural networks on resource-constrained devices, but finding the optimal bit-width for each layer represents a complex combinatorial optimization problem. Current state-of-the-art methods rely on computationally expensive search algorithms or local sensitivity heuristic proxies like the Hessian, which fail to capture the cascading global effects of quantization error. In this work, we argue that the quantization sensitivity of a layer should not be measured by its local properties, but by its impact on the information flow throughout the entire network. We introduce InfoQ, a novel framework for MPQ that is training-free in the bit-width search phase. InfoQ assesses layer sensitivity by quantizing each layer at different bit-widths and measuring, through a single forward pass, the resulting change in mutual information in the subsequent layers. This quantifies how much each layer quantization impacts the network information flow. The resulting scores are used to formulate bit-width allocation as an integer linear programming problem, which is solved efficiently to minimize total sensitivity under a given budget (e.g., model size or BitOps). Our retraining-free search phase provides a superior search-time/accuracy trade-off (using two orders of magnitude less data compared to state-of-the-art methods such as LIMPQ), while yielding up to a 1% accuracy improvement for MobileNetV2 and ResNet18 on ImageNet at high compression rates (14X and 10.66X).

cs.LG↗

arXiv2024

Content Aware Analysis of Scholarly Networks: A Case Study on CORD19 Dataset

This paper investigates the relationships among key elements of the scientific research network, namely articles, researchers, and journals. We introduce a novel approach to use semantic information through the HITS algorithm-based propagation of topic information in the network. The topic information is derived by using the Named Entity Recognition and Entity Linkage. In our case, MedCAT is used to extract the topics from the CORD19 Dataset, which is a corpus of academic articles about COVID-19 and the coronavirus scientific network. Our approach focuses on the COVID-19 domain, utilizing the CORD-19 dataset to demonstrate the efficacy of integrating topic-related information within the citation framework. Through the application of a hybrid HITS algorithm, we show that incorporating topic data significantly influences article rankings, revealing deeper insights into the structure of the academic community.

cs.SI↗