Ocean Monjur

Tadesse Destaw Belay, Shahriar Kabir Nahin, Israel Abebe Azime et al.

How can language learning systems be developed for languages that lack sufficient training resources? This challenge is increasingly faced by developers across the African continent who aim to build AI systems capable of understanding and responding in local languages. To address this gap, we introduce AFRILANGDICT, a collection of 194.7K African language-English dictionary entries designed as seed resources for generating language-learning materials, enabling us to automatically construct large-scale, diverse, and verifiable student-tutor question-answer interactions suitable for training AI-assisted language tutors. Using AFRILANGDICT, we build AFRILANGEDU, a dataset of 78.9K multi-turn training examples for Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO). Using AFRILANGEDU, we train language tutoring models collectively referred to as AFRILANGTUTOR. We fine-tune two multilingual LLMs: Llama-3-8B-IT and Gemma-3-12B-IT on AFRILANGEDU across 10 African languages and evaluate their performance. Our results show that models trained on AFRILANGEDU consistently outperform their base counterparts, and combining SFT and DPO yields substantial improvements, with gains ranging from 1.8% to 15.5% under LLM-as-a-judge evaluations across four criteria. To facilitate further research on low-resource languages -- all resources are available at https://huggingface.co/afrilang-edu.

Ocean Monjur, Shahriar Kabir Nahin, Anshuman Chhabra

While current Large Language Models (LLMs) exhibit remarkable reasoning capabilities through test-time compute scaling (TTS), their massive parameter counts and high inference costs have motivated the development of pruning methods that can reduce model size without sacrificing performance. However, specific to reasoning LLMs, prior work has shown that structured pruning (methods which removes entire set of layer blocks), significantly degrades TTS reasoning performance. In this work, we revisit this assumption and instead investigate whether unstructured pruning (methods that carefully remove only certain redundant/detrimental weights) exhibits similar limitations. Surprisingly, our extensive experiments across four reasoning benchmarks on two reasoning LLMs: s1.1-7B and Qwen3-8B, consistently show that unstructured pruning augments TTS performance compared to structured pruning, and at times can even outperform the unpruned full-weight LLMs. Furthermore, we also empirically study the impact of different layer-wise sparsity allocation strategies, which are an important parametric choice for instantiating unstructured pruning methods. These findings challenge the conventional notion that pruning always reduces TTS performance and in fact, suggest that carefully undertaken pruning can improve TTS effectiveness even further.

Md. Toukir Ahmed, Ocean Monjur, Mohammed Kamruzzaman

Hyperspectral imaging (HSI) has recently emerged as a promising tool for many agricultural applications; however, the technology cannot be directly used in a real-time system due to the extensive time needed to process large volumes of data. Consequently, the development of a simple, compact, and cost-effective imaging system is not possible with the current HSI systems. Therefore, the overall goal of this study was to reconstruct hyperspectral images from RGB images through deep learning for agricultural applications. Specifically, this study used Hyperspectral Convolutional Neural Network - Dense (HSCNN-D) to reconstruct hyperspectral images from RGB images for predicting soluble solid content (SSC) in sweet potatoes. The algorithm accurately reconstructed the hyperspectral images from RGB images, with the resulting spectra closely matching the ground-truth. The partial least squares regression (PLSR) model based on reconstructed spectra outperformed the model using the full spectral range, demonstrating its potential for SSC prediction in sweet potatoes. These findings highlight the potential of deep learning-based hyperspectral image reconstruction as a low-cost, efficient tool for various agricultural uses.

Md. Toukir Ahmed, Md Wadud Ahmed, Ocean Monjur et al.

As the demand for food surges and the agricultural sector undergoes a transformative shift towards sustainability and efficiency, the need for precise and proactive measures to ensure the health and welfare of livestock becomes paramount. In the context of the broader agricultural landscape outlined, the application of Hyperspectral Imaging (HSI) takes on profound significance. HSI has emerged as a cutting-edge, non-destructive technique for fast and accurate egg quality analysis, including the detection of chick embryo mortality. However, the high cost and operational complexity compared to conventional RGB imaging are significant bottlenecks in the widespread adoption of HSI technology. To overcome these hurdles and unlock the full potential of HSI, a promising solution is hyperspectral image reconstruction from standard RGB images. This study aims to reconstruct hyperspectral images from RGB images for non-destructive early prediction of chick embryo mortality. Firstly, the performance of different image reconstruction algorithms, such as HRNET, MST++, Restormer, and EDSR were compared to reconstruct the hyperspectral images of the eggs in the early incubation period. Later, the reconstructed spectra were used to differentiate live from dead chick-producing eggs using the XGBoost and Random Forest classification methods. Among the reconstruction methods, HRNET showed impressive reconstruction performance with MRAE of 0.0955, RMSE of 0.0159, and PSNR of 36.79 dB. This study motivated that harnessing imaging technology integrated with smart sensors and data analytics has the potential to improve automation, enhance biosecurity, and optimize resource management towards sustainable agriculture 4.0.

Mk Bashar, Ocean Monjur, Samia Islam et al.

In recent years, Convolutional Neural Networks (CNNs), MLP-mixers, and Vision Transformers have risen to prominence as leading neural architectures in image classification. Prior research has underscored the distinct advantages of each architecture, and there is growing evidence that combining modules from different architectures can boost performance. In this study, we build upon and improve previous work exploring the complementarity between different architectures. Instead of heuristically merging modules from various architectures through trial and error, we preserve the integrity of each architecture and combine them using ensemble techniques. By maintaining the distinctiveness of each architecture, we aim to explore their inherent complementarity more deeply and with implicit isolation. This approach provides a more systematic understanding of their individual strengths. In addition to uncovering insights into architectural complementarity, we showcase the effectiveness of even basic ensemble methods that combine models from diverse architectures. These methods outperform ensembles comprised of similar architectures. Our straightforward ensemble framework serves as a foundational strategy for blending complementary architectures, offering a solid starting point for further investigations into the unique strengths and synergies among different architectures and their ensembles in image classification. A direct outcome of this work is the creation of an ensemble of classification networks that surpasses the accuracy of the previous state-of-the-art single classification network on ImageNet, setting a new benchmark, all while requiring less overall latency.

Abdullah Bin Shams, Md. Mohsin Sarker Raihan, Md. Mohi Uddin Khan et al.

To contain the spread of the virus and stop the overcrowding of hospitalized patients, the coronavirus pandemic crippled healthcare facilities, mandating lockdowns and promoting remote work. As a result, telehealth has become increasingly popular for offering low-risk care to patients. However, the difficulty of preventing the next potential waves of infection has increased by constant virus mutation into new forms and a general lack of test kits, particularly in developing nations. In this research, a unique cloud-based application for the early identification of individuals who may have COVID-19 infection is proposed. The application provides five modes of diagnosis from possible symptoms (f1), cough sound (f2), specific blood biomarkers (f3), Raman spectral data of blood specimens (f4), and ECG signal paper-based image (f5). When a user selects an option and enters the information, the data is sent to the cloud server. The deployed machine learning (ML) and deep learning (DL) models classify the data in real time and inform the user of the likelihood of COVID-19 infection. Our deployed models can classify with an accuracy of 100%, 99.80%, 99.55%, 95.65%, and 77.59% from f3, f4, f5, f2, and f1 respectively. Moreover, the sensitivity for f2, f3, and f4 is 100%, which indicates the correct identification of COVID positive patients. This is significant in limiting the spread of the virus. Additionally, another ML model, as seen to offer 92% accuracy serves to identify patients who, out of a large group of patients admitted to the hospital cohort, need immediate critical care support by estimating the mortality risk of patients from blood parameters. The instantaneous multimodal nature of our technique offers multiplex and accurate diagnostic methods, highlighting the effectiveness of telehealth as a simple, widely available, and low-cost diagnostic solution, even for future pandemics.

S. M Mehedi Zaman, Wasay Mahmood Qureshi, Md. Mohsin Sarker Raihan et al.

Cardiovascular disease, especially heart failure is one of the major health hazard issues of our time and is a leading cause of death worldwide. Advancement in data mining techniques using machine learning (ML) models is paving promising prediction approaches. Data mining is the process of converting massive volumes of raw data created by the healthcare institutions into meaningful information that can aid in making predictions and crucial decisions. Collecting various follow-up data from patients who have had heart failures, analyzing those data, and utilizing several ML models to predict the survival possibility of cardiovascular patients is the key aim of this study. Due to the imbalance of the classes in the dataset, Synthetic Minority Oversampling Technique (SMOTE) has been implemented. Two unsupervised models (K-Means and Fuzzy C-Means clustering) and three supervised classifiers (Random Forest, XGBoost and Decision Tree) have been used in our study. After thorough investigation, our results demonstrate a superior performance of the supervised ML algorithms over unsupervised models. Moreover, we designed and propose a supervised stacked ensemble learning model that can achieve an accuracy, precision, recall and F1 score of 99.98%. Our study shows that only certain attributes collected from the patients are imperative to successfully predict the surviving possibility post heart failure, using supervised ML algorithms.