50 research outputs found
Comparison Effect on Biogas Production from Vegetable and Fruit Waste with Rumen Digesta Through Co-Digestion Process
No full textBiogas is the best renewable energy as it can be produced from any biomass for example any plant or living organism. The purpose of this research was to produce biomethane from co-digestion of vegetable and fruit waste with rumen digesta through anaerobic digestion process. In this research, two trials of experiment were conducted. Each trial has three different sample with different mixing ratios. Raw materials used in the experiment was rumen digesta of goat and cow, potato, capsicum, cucumbers, onions, radish, cauliflower, carrot, leafy vegetables, apple, banana, and papaya. In each sample, 1200 gram of raw materials were used. Hydraulic retention time was 30 days. Data was collected by water displacement method. The experiment found that the gas production started from 2nd or 3rd days and stops in 28th or 29th day. Highest production of biogas was 35, 33, 30, 40, 50 and 35 mL/day on the 17th, 14th, 17th, 11th, 12th and 7th day at the mixing ratios of 1:1:2, 1:2:1, 1:1.5:1.5, 1:0.5:0.5, 1:2:2 and 1.5:1.5:1 (Rumen Digesta: Vegetable Waste: Fruit Waste) respectively. The study suggests making digester for the recycling of waste to produce biogas, a renewable and environment friendly energy.Tasnim, Anika; Mamun, Muhammad; Hossen, Md Anwar; Rahman, Towfiq. (2022). Comparison Effect on Biogas Production from Vegetable and Fruit Waste with Rumen Digesta Through Co-Digestion Process. Retrieved from the University Digital Conservancy, 10.24018/ejenergy.2022.2.1.38
Effect of urea fertilizer deep placement days after transplanting using brri prilled urea applicator on transplanted boro rice yield
No full textRice is the most important crop in the developing countries of Asia. In the south and south-east Asia, rain-fed and irrigated transplanted rice occupies nearly two-thirds of the rice-growing area and produces more than 80% of the rough rice. In these areas, prilled urea conventionally applied by farmers is very insufficient in the transplanted rice field, where severe losses occur (up to 60% of applied N) via NH3volatilization, denitrification, leaching, and runoff. Considering loss minimization, an experiment was conducted during the Boro season at Bangladesh Agricultural Development Corporation (BADC) in Sylhet to evaluate the performance of BRRI Prilled Urea Applicator (BPUA) at the different periods after transplanting BRRI dhan28. The results reveal that the field performance of the BPUA was suitable on first day after seedling transplanting under sandy clay loam soil compared to the third day after transplanting (DAT). At the 105 DAT, the height of the crop was found to be 104.3, 104.3, and 95.7 cm for urea deep placement by BPUA on first, second, and third day after seedling transplanting respectively. The maximum grain and straw yield was found at 6.8 t ha-1 and 5.2 t ha-1, respectively which varied with the date of applicator operation after seedling transplanting. The benefit-cost ratio was found 1.63 at first DAT, whereas it was lower on the third days after seedling transplanting. Farmer can apply urea fertilizer in the non-oxidized zone by the BPUA after the first and second day of seedling transplanting in the sandy clay loam soil for maximum yield.Mamun, Muhammad; Nahar, K.; Rahman, Towfiq; Hossen, Md Anwar. (2020). Effect of urea fertilizer deep placement days after transplanting using brri prilled urea applicator on transplanted boro rice yield. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/276908
FedProtoKD: Dual Knowledge Distillation with Adaptive Class-wise Prototype Margin for Heterogeneous Federated Learning
No full textHeterogeneous Federated Learning (HFL) has gained attention for its ability to accommodate diverse models and heterogeneous data across clients. Prototype-based HFL methods emerge as a promising solution to address statistical heterogeneity and privacy challenges, paving the way for new advancements in HFL research. This method focuses on sharing only class-representative prototypes among heterogeneous clients. However, these prototypes are often aggregated on the server using weighted averaging, leading to sub-optimal global knowledge; these cause the shrinking of aggregated prototypes, which negatively affects the model performance in scenarios when models are heterogeneous and data distributions are extremely non-IID. We propose FedProtoKD in a Heterogeneous Federated Learning setting, using an enhanced dual-knowledge distillation mechanism to improve the system performance with clients' logits and prototype feature representation. We aim to resolve the prototype margin-shrinking problem using a contrastive learning-based trainable server prototype by leveraging a class-wise adaptive prototype margin. Furthermore, we assess the importance of public samples using the closeness of the sample's prototype to its class representative prototypes, which enhances learning performance. FedProtoKD achieved average improvements of 1.13% up to 34.13% accuracy across various settings and significantly outperforms existing state-of-the-art HFL methods.This is a preprint from Hossen, Md Anwar, Fatema Siddika, Wensheng Zhang, Anuj Sharma, and Ali Jannesari. "FedProtoKD: Dual Knowledge Distillation with Adaptive Class-wise Prototype Margin for Heterogeneous Federated Learning." arXiv preprint arXiv:2508.19009 (2025). doi: https://doi.org/10.48550/arXiv.2508.19009
FedReFT: Federated Representation Fine-Tuning with All-But-Me Aggregation
No full textParameter-efficient fine-tuning (PEFT) has attracted significant attention for adapting large pre-trained models by modifying a small subset of parameters. Recently, Representation Fine-tuning (ReFT) has emerged as an effective alternative. ReFT shifts the fine-tuning paradigm from updating model weights to directly manipulating hidden representations that capture rich semantic information, and performs better than state-of-the-art PEFTs in standalone settings. However, its application in Federated Learning (FL) remains challenging due to heterogeneity in clients' data distributions, model capacities, and computational resources. To address these challenges, we introduce Federated Representation Fine-Tuning (FedReFT), a novel approach to fine-tune the client's hidden representation. FedReFT applies sparse intervention layers to steer hidden representations directly, offering a lightweight and semantically rich fine-tuning alternative ideal for edge devices. However, representation-level updates are especially vulnerable to aggregation mismatch under different task heterogeneity, where naive averaging can corrupt semantic alignment. To mitigate this issue, we propose All-But-Me (ABM) aggregation, where each client receives the aggregated updates of others and partially incorporates them, enabling stable and personalized learning by balancing local focus with global knowledge. We evaluate FedReFT on commonsense reasoning, arithmetic reasoning, instruction-tuning, and GLUE, where it consistently outperforms state-of-the-art PEFT methods in FL, achieving 7x-15x higher parameter efficiency compared to leading LoRA-based approaches.This is a preprint from Siddika, Fatema, Md Anwar Hossen, J. Pablo Muñoz, Tanya Roosta, Anuj Sharma, and Ali Jannesari. "FedReFT: Federated Representation Fine-Tuning with All-But-Me Aggregation." arXiv preprint arXiv:2508.20295 (2025). doi: https://doi.org/10.48550/arXiv.2508.20295
Fair allocation of bandwidth at edge servers for concurrent federated learning processes
No full textCloud servers can handle large amounts of data and many devices, but they have slow and unstable communication over extended networks. Edge servers work with smaller data, but have faster and more stable communication with nearby devices. This led to the client-edge-cloud system. As FL grows, more processes need a set-up with multiple FL servers. Edge servers have limited bandwidth and must be shared between FL servers and clients, limiting how many requests they can handle simultaneously. This work explores concurrent FL processes within a three-tier system, with edge servers between edge devices and FL servers. A challenge in this setup is the limited bandwidth from edge devices to edge servers. Thus, allocating the bandwidth efficiently and fairly to support simultaneous FL processes becomes crucial. We propose a game-theoretic approach to model the bandwidth allocation problem and develop distributed and centralized heuristic schemes to find an approximate Nash equilibrium of the game. Through rigorous analysis and experimentation, we demonstrate that our schemes efficiently and fairly assign the bandwidth to the FL processes and outperform a baseline scheme where each edge server assigns bandwidth proportionally to the FL servers’ requests that it receives. The proposed distributed and centralized schemes have similar performance
Environmentally Sustainable Tourism Problem and Prospects in Bangladesh: A Study on Coastal Region at Cox’s Bazar
No full textCoastal region in Bangladesh is always includes floodplains mangroves marshes and fringing coral reefs. Tourism is considered as the world’s largest and rapid growing industry of modern business world. But tourism industry is directly related to the environment. The quality of the environment, both natural and man-made is essential to tourism. The negative impacts of tourism development can gradually destroy environmental resources like loss of marine resources due to destruction of coral reefs overfishing pollution of marine and freshwater resources soil degradation and loss of land resources, air pollution, natural hazards and sea level rise and climate change. Environmental Education must be nurtured from root level to the last day of life to provide the best fruit of sustainable development. This study focuses on the guidelines of the way and contents of different level of environmental Education. Environmentally educated society can carry out the sustainable tourism development of coastal Bangladesh. Keywords: Sustainable Tourism Development, Sustainable Coastal Zone, Eco-tourism, Coastal Tourism., Environmental Impacts
Problems and Possibilities of Good Governance in Bangladesh: Hurdles to Achieve Sustainable Development
No full textConsidering Bangladesh's history of tumultuous politics, the country is gradually improving in terms of political stability and governance. Unfortunately, development has not been stable, mainly due to political will and commitment. Democracy and governance in Bangladesh, still plagued by violence, corruption, outdated laws, abuse of human rights, absence of rule of law, non accountability, and heavy politicization of all government institutions including the judicial system. It is observed that we need strong political leadership with commitment to fight against deep rooted corruption, non accountability, non transparency and inefficiency is imperative for establishing good governance and to ensure sustainable development. Keeping all these issues in consideration, this research paper discusses governance and good governance related issues and sustainable development. It also provides a set of recommendations based on this analysis. The study is explanatory in nature which is based on extensive literature review and secondary sources. Keywords: Governance, Good Governance, Sustainable Development, Transparent and Efficient, Politicization, Accountability, Political will and Commitment, Outdated Laws
Fair Allocation of Bandwidth At Edge Servers For Concurrent Hierarchical Federated Learning
No full textThis paper explores concurrent FL processes within a three-tier system, with edge servers between edge devices and FL servers. A challenge in this setup is the limited bandwidth from edge devices to edge servers. Thus, allocating the bandwidth efficiently and fairly to support simultaneous FL processes becomes crucial. We propose a game-theoretic approach to model the bandwidth allocation problem and develop distributed and centralized heuristic schemes to find an approximate Nash Equilibrium of the game. We proposed the approach mentioned above using centralized and entirely distributed assumptions. Through rigorous analysis and experimentation, we demonstrate that our schemes efficiently and fairly assign the bandwidth to the FL processes for centralized and distributed solutions and outperform a baseline scheme where each edge server assigns bandwidth proportionally to the FL servers\u27 requests that it receives. The proposed distributed and centralized schemes have comptetive performance
Exploring the possibility of using Agroplus Biodecomposer for boosting up rice productivity under Bangladesh condition
No full textOver dependence on chemical fertilizers is a threat to the sustainability of rice ecosystem. Application of organic and biofertilizers might reduce reliance on chemical fertilizers and thus can play a vital role to boost up rice productivity in an eco-friendly way. An experiment was conducted at Mymensingh (24°10'0'' N latitude and 90°25'0" E longitude at 15 m above the sea level), Bangladesh during November 2015 to April 2016 to evaluate the effect of different dosages of Agroplus Biodecomposer, an organic biofertilizer containing Streptomycetes bacteria, on the growth and yield performance of some rice. The experiment included four winter rice varieties viz. (i) Hybrid rice Hira and (ii) Hybrid rice Tej (iii) BRRI dhan28 and (iv) BRRI dhan29; and four concentrations of Agroplus Biodecomposer viz. (i) no Agroplus Biodecomposer (Control), (ii) 2% Agroplus Biodecomposer (iii) 3% Agroplus Biodecomposer and (iv) 4% Agroplus Biodecomposer. The experiment was laid out in a randomized complete block design (RCBD) with three replications. Agroplus Biodecomposer positively influenced growth and productivity of winter rice. It was evident that both plant height and tillering ability of winter rice were increased gradually with increased concentration of Agroplus Biodecomposer at all the growth stages of rice. All the yield contributing characters of rice were enhanced due to Agroplus Biodecomposer application which resulted in increased grain yield. Compared to control, rice grain yield was increased by 14, 20 and 28%, respectively due to application of Agroplus Biodecomposer at 2, 3 and 4% concentration. Rice variety also differed significantly in terms of growth and yield performance among themselves. Hybrid varieties performed better than inbred ones. Hybrid variety Hira appeared as the best performer followed by another hybrid Tej. Hybrid variety Hira interacted favorably with 4% Agroplus Biodecomposer to produce the highest grain yield of rice (7 t ha-1). Therefore, biofertilizer Agroplus Biodecomposer can be introduced to Bangladesh and foliar spray with 4% Agroplus Biodecomposer at 30, 45 and 60 DAT can be recommended for boosting up winter rice yield in a sustainable way. [Fundam Appl Agric 2018; 3(1.000): 372-381
