Mohammad Ariful Haque

Ismam Nur Swapnil, Aranya Saha, Tanvir Ahmed Khan et al.

Reinforcement learning is widely used to improve the reasoning ability of large language models, especially when answers can be automatically checked. Standard GRPO-style training updates the model using only the current step, while full multi-step lookahead can give a better update direction but is too expensive because it needs many backward passes. We propose Gradient Extrapolation-Based Policy Optimization (GXPO), a plug-compatible policy-update rule for GRPO-style reasoning RL. GXPO approximates a longer local lookahead using only three backward passes during an active phase. It reuses the same batch of rollouts, rewards, advantages, and GRPO loss, so it does not require new rollouts or reward computation at the lookahead points. GXPO takes two fast optimizer steps, measures how the gradients change, predicts a virtual K-step lookahead point, moves the policy partway toward that point, and then applies a corrective update using the true gradient at the new position. When the lookahead signal becomes unstable, GXPO automatically switches back to standard single-pass GRPO. We also give a plain-gradient-descent surrogate analysis that explains when the extrapolation is exact and where its local errors come from. Across Qwen2.5 and Llama math-reasoning experiments, GXPO improves the average sampled pass@1 by +1.65 to +5.00 points over GRPO and by +0.14 to +1.28 points over the strongest SFPO setting, while keeping the active-phase cost fixed at three backward passes. It also achieves up to 4.00x step speedup, 2.33x wall-clock speedup, and 1.33x backward-pass speedup in reaching GRPO's peak accuracy.

Ashik Abrar Naeem, Mohammad Ariful Haque

Autonomous navigation and obstacle avoidance remain a core challenge of modern Unmanned Aerial Vehicles (UAVs). While traditional control methods struggle with the complexity and variability of the environment, reinforcement learning (RL) enables UAVs to learn adaptive behaviors through interaction with the environment. Existing research with RL prioritizes the mission success at the expense of mission time and safety of UAVs. This study integrates Potential Based Reward Shaping (PBRS) with Control Lyapunov Functions (CLF) and Control Barrier Functions (CBF) to simultaneously optimize mission time and ensure formal safety guarantees. An RL model is trained in a generalized simple environment, then used in complex scenarios incorporating a CLF-CBF-QP filter without further training. Experimental results in simulated environments demonstrate a significant reduction in mission time and outstanding performance in complex environment.

Ismam Nur Swapnil, Aranya Saha, Tanvir Ahmed Khan et al.

Vision-Language Models (VLMs) show promise in medical image analysis, yet their capacity for structured reasoning in complex domains like dermatology is often limited by data scarcity and the high computational cost of advanced training techniques. To address these challenges, we introduce DermIQ-VLM, a VLM developed through a multi-stage, resource-efficient methodology designed to emulate a dermatologist's diagnostic process. Our primary contribution is a modified version of Grouped Relative Policy Optimization (GRPO), called GRPO++, which stabilizes the powerful but data-intensive GRPO framework. Our proposed training pipeline first employs GRPO++ for reasoning-oriented disease recognition, followed by supervised fine-tuning for conversational ability. To mitigate factual errors introduced during this step, we then align the model using Direct Preference Optimization (DPO), leveraging a Knowledge Graph-based system as a scalable proxy for expert preference. A preliminary evaluation on a curated dermatological dataset demonstrates that our proposed methodology yields notable performance gains over standard fine-tuning approaches. These findings validate the potential of our pipeline as a feasible pathway for developing specialized, reliable VLMs in resource-constrained environments.

Aranya Saha, Tanvir Ahmed Khan, Ismam Nur Swapnil et al.

Vision-language models (VLMs) have shown significant potential for medical tasks; however, their general-purpose nature can limit specialized diagnostic accuracy, and their large size poses substantial inference costs for real-world clinical deployment. To address these challenges, we introduce CLARIFY, a Specialist-Generalist framework for dermatological visual question answering (VQA). CLARIFY combines two components: (i) a lightweight, domain-trained image classifier (the Specialist) that provides fast and highly accurate diagnostic predictions, and (ii) a powerful yet compressed conversational VLM (the Generalist) that generates natural language explanations to user queries. In our framework, the Specialist's predictions directly guide the Generalist's reasoning, focusing it on the correct diagnostic path. This synergy is further enhanced by a knowledge graph-based retrieval module, which grounds the Generalist's responses in factual dermatological knowledge, ensuring both accuracy and reliability. This hierarchical design not only reduces diagnostic errors but also significantly improves computational efficiency. Experiments on our curated multimodal dermatology dataset demonstrate that CLARIFY achieves an 18\% improvement in diagnostic accuracy over the strongest baseline, a fine-tuned, uncompressed single-line VLM, while reducing the average VRAM requirement and latency by at least 20\% and 5\%, respectively. These results indicate that a Specialist-Generalist system provides a practical and powerful paradigm for building lightweight, trustworthy, and clinically viable AI systems.

Shadman Sobhan, Mohammad Ariful Haque

Large Language Models (LLMs) are capable of natural language understanding and generation. But they face challenges such as hallucination and outdated knowledge. Fine-tuning is one possible solution, but it is resource-intensive and must be repeated with every data update. Retrieval-Augmented Generation (RAG) offers an efficient solution by allowing LLMs to access external knowledge sources. However, traditional RAG pipelines struggle with retrieving information from complex technical documents with structured data such as tables and images. In this work, we propose a RAG pipeline, capable of handling tables and images in documents, for technical documents that support both scanned and searchable formats. Its retrieval process combines vector similarity search with a fine-tuned reranker based on Gemma-2-9b-it. The reranker is trained using RAFT (Retrieval-Augmented Fine-Tuning) on a custom dataset designed to improve context identification for question answering. Our evaluation demonstrates that the proposed pipeline achieves a high faithfulness score of 94% (RAGas) and 96% (DeepEval), and an answer relevancy score of 87% (RAGas) and 93% (DeepEval). Comparative analysis demonstrates that the proposed architecture is superior to general RAG pipelines in terms of table-based questions and handling questions outside context.

Parnian Afshar, Arash Mohammadi, Konstantinos N. Plataniotis et al.

Lung cancer is one of the deadliest cancers, and in part its effective diagnosis and treatment depend on the accurate delineation of the tumor. Human-centered segmentation, which is currently the most common approach, is subject to inter-observer variability, and is also time-consuming, considering the fact that only experts are capable of providing annotations. Automatic and semi-automatic tumor segmentation methods have recently shown promising results. However, as different researchers have validated their algorithms using various datasets and performance metrics, reliably evaluating these methods is still an open challenge. The goal of the Lung-Originated Tumor Segmentation from Computed Tomography Scan (LOTUS) Benchmark created through 2018 IEEE Video and Image Processing (VIP) Cup competition, is to provide a unique dataset and pre-defined metrics, so that different researchers can develop and evaluate their methods in a unified fashion. The 2018 VIP Cup started with a global engagement from 42 countries to access the competition data. At the registration stage, there were 129 members clustered into 28 teams from 10 countries, out of which 9 teams made it to the final stage and 6 teams successfully completed all the required tasks. In a nutshell, all the algorithms proposed during the competition, are based on deep learning models combined with a false positive reduction technique. Methods developed by the three finalists show promising results in tumor segmentation, however, more effort should be put into reducing the false positive rate. This competition manuscript presents an overview of the VIP-Cup challenge, along with the proposed algorithms and results.

Jayanta Dey, Mohammad Ariful Haque

Information regarding the location of power distribution grid can be extracted from the power signature embedded in the multimedia signals (e.g., audio, video data) recorded near electrical activities. This implicit mechanism of identifying the origin-of-recording can be a very promising tool for multimedia forensics and security applications. In this work, we have developed a novel grid-of-origin identification system from media recording that consists of a number of support vector machine (SVM) followed by pole-matching (PM) classifiers. First, we determine the nominal frequency of the grid (50 or 60 Hz) based on the spectral observation. Then an SVM classifier, trained for the detection of a grid with a particular nominal frequency, narrows down the list of possible grids on the basis of different discriminating features extracted from the electric network frequency (ENF) signal. The decision of the SVM classifier is then passed to the PM classifier that detects the final grid based on the minimum distance between the estimated poles of test and training grids. Thus, we start from the problem of classifying grids with different nominal frequencies and simplify the problem of classification in three stages based on nominal frequency, SVM and finally using PM classifier. This cascaded system of classification ensures better accuracy (15.57% higher) compared to traditional ENF-based SVM classifiers described in the literature.

Jayanta Dey, Md Sanzid Bin Hossain, Mohammad Ariful Haque

Voice activity detection (VAD), used as the front end of speech enhancement, speech and speaker recognition algorithms, determines the overall accuracy and efficiency of the algorithms. Therefore, a VAD with low complexity and high accuracy is highly desirable for speech processing applications. In this paper, we propose a novel training method on large dataset for supervised learning-based VAD system using support vector machine (SVM). Despite of high classification accuracy of support vector machines (SVM), trivial SVM is not suitable for classification of large data sets needed for a good VAD system because of high training complexity. To overcome this problem, a novel ensemble-based approach using SVM has been proposed in this paper.The performance of the proposed ensemble structure has been compared with a feedforward neural network (NN). Although NN performs better than single SVM-based VAD trained on a small portion of the training data, ensemble SVM gives accuracy comparable to neural network-based VAD. Ensemble SVM and NN give 88.74% and 86.28% accuracy respectively whereas the stand-alone SVM shows 57.05% accuracy on average on the test dataset.

Md. Shamim Hussain, Mohammad Ariful Haque

Speech, Music and Noise classification/segmentation is an important preprocessing step for audio processing/indexing. To this end, we propose a novel 1D Convolutional Neural Network (CNN) - SwishNet. It is a fast and lightweight architecture that operates on MFCC features which is suitable to be added to the front-end of an audio processing pipeline. We showed that the performance of our network can be improved by distilling knowledge from a 2D CNN, pretrained on ImageNet. We investigated the performance of our network on the MUSAN corpus - an openly available comprehensive collection of noise, music and speech samples, suitable for deep learning. The proposed network achieved high overall accuracy in clip (length of 0.5-2s) classification (>97% accuracy) and frame-wise segmentation (>93% accuracy) tasks with even higher accuracy (>99%) in speech/non-speech discrimination task. To verify the robustness of our model, we trained it on MUSAN and evaluated it on a different corpus - GTZAN and found good accuracy with very little fine-tuning. We also demonstrated that our model is fast on both CPU and GPU, consumes a low amount of memory and is suitable for implementation in embedded systems.

Ahmed Nazim Uddin, Md Ashequr Rahman, Md. Rafidul Islam et al.

The task of determining a speaker's native language based only on his speeches in a second language is known as Native Language Identification or NLI. Due to its increasing applications in various domains of speech signal processing, this has emerged as an important research area in recent times. In this paper we have proposed an i-vector based approach to develop an automatic NLI system using MFCC and GFCC features. For evaluation of our approach, we have tested our framework on the 2016 ComParE Native language sub-challenge dataset which has English language speakers from 11 different native language backgrounds. Our proposed method outperforms the baseline system with an improvement in accuracy by 21.95% for the MFCC feature based i-vector framework and 22.81% for the GFCC feature based i-vector framework.