-
Adaptive Tabu Dropout for Regularization of Deep Neural Network
Authors:
Md. Tarek Hasan,
Arifa Akter,
Mohammad Nazmush Shamael,
Md Al Emran Hossain,
H. M. Mutasim Billah,
Sumayra Islam,
Swakkhar Shatabda
Abstract:
Dropout is an effective strategy for the regularization of deep neural networks. Applying tabu to the units that have been dropped in the recent epoch and retaining them for training ensures diversification in dropout. In this paper, we improve the Tabu Dropout mechanism for training deep neural networks in two ways. Firstly, we propose to use tabu tenure, or the number of epochs a particular unit…
▽ More
Dropout is an effective strategy for the regularization of deep neural networks. Applying tabu to the units that have been dropped in the recent epoch and retaining them for training ensures diversification in dropout. In this paper, we improve the Tabu Dropout mechanism for training deep neural networks in two ways. Firstly, we propose to use tabu tenure, or the number of epochs a particular unit will not be dropped. Different tabu tenures provide diversification to boost the training of deep neural networks based on the search landscape. Secondly, we propose an adaptive tabu algorithm that automatically selects the tabu tenure based on the training performances through epochs. On several standard benchmark datasets, the experimental results show that the adaptive tabu dropout and tabu tenure dropout diversify and perform significantly better compared to the standard dropout and basic tabu dropout mechanisms.
△ Less
Submitted 31 December, 2024;
originally announced January 2025.
-
Deep Transfer Learning Based Peer Review Aggregation and Meta-review Generation for Scientific Articles
Authors:
Md. Tarek Hasan,
Mohammad Nazmush Shamael,
H. M. Mutasim Billah,
Arifa Akter,
Md Al Emran Hossain,
Sumayra Islam,
Salekul Islam,
Swakkhar Shatabda
Abstract:
Peer review is the quality assessment of a manuscript by one or more peer experts. Papers are submitted by the authors to scientific venues, and these papers must be reviewed by peers or other authors. The meta-reviewers then gather the peer reviews, assess them, and create a meta-review and decision for each manuscript. As the number of papers submitted to these venues has grown in recent years,…
▽ More
Peer review is the quality assessment of a manuscript by one or more peer experts. Papers are submitted by the authors to scientific venues, and these papers must be reviewed by peers or other authors. The meta-reviewers then gather the peer reviews, assess them, and create a meta-review and decision for each manuscript. As the number of papers submitted to these venues has grown in recent years, it becomes increasingly challenging for meta-reviewers to collect these peer evaluations on time while still maintaining the quality that is the primary goal of meta-review creation. In this paper, we address two peer review aggregation challenges a meta-reviewer faces: paper acceptance decision-making and meta-review generation. Firstly, we propose to automate the process of acceptance decision prediction by applying traditional machine learning algorithms. We use pre-trained word embedding techniques BERT to process the reviews written in natural language text. For the meta-review generation, we propose a transfer learning model based on the T5 model. Experimental results show that BERT is more effective than the other word embedding techniques, and the recommendation score is an important feature for the acceptance decision prediction. In addition, we figure out that fine-tuned T5 outperforms other inference models. Our proposed system takes peer reviews and other relevant features as input to produce a meta-review and make a judgment on whether or not the paper should be accepted. In addition, experimental results show that the acceptance decision prediction system of our task outperforms the existing models, and the meta-review generation task shows significantly improved scores compared to the existing models. For the statistical test, we utilize the Wilcoxon signed-rank test to assess whether there is a statistically significant improvement between paired observations.
△ Less
Submitted 5 October, 2024;
originally announced October 2024.
-
An Artificial Intelligence-based Framework to Achieve the Sustainable Development Goals in the Context of Bangladesh
Authors:
Md. Tarek Hasan,
Mohammad Nazmush Shamael,
Arifa Akter,
Rokibul Islam,
Md. Saddam Hossain Mukta,
Salekul Islam
Abstract:
Sustainable development is a framework for achieving human development goals. It provides natural systems' ability to deliver natural resources and ecosystem services. Sustainable development is crucial for the economy and society. Artificial intelligence (AI) has attracted increasing attention in recent years, with the potential to have a positive influence across many domains. AI is a commonly e…
▽ More
Sustainable development is a framework for achieving human development goals. It provides natural systems' ability to deliver natural resources and ecosystem services. Sustainable development is crucial for the economy and society. Artificial intelligence (AI) has attracted increasing attention in recent years, with the potential to have a positive influence across many domains. AI is a commonly employed component in the quest for long-term sustainability. In this study, we explore the impact of AI on three pillars of sustainable development: society, environment, and economy, as well as numerous case studies from which we may deduce the impact of AI in a variety of areas, i.e., agriculture, classifying waste, smart water management, and Heating, Ventilation, and Air Conditioning (HVAC) systems. Furthermore, we present AI-based strategies for achieving Sustainable Development Goals (SDGs) which are effective for developing countries like Bangladesh. The framework that we propose may reduce the negative impact of AI and promote the proactiveness of this technology.
△ Less
Submitted 23 April, 2023;
originally announced April 2023.
-
Finding the Contextual Gap Towards Employee Engagement in Financial Sector: A Review Study
Authors:
Habiba Akter,
Ilham Sentosa,
Sheikh Muhamad Hizam,
Waqas Ahmed,
Arifa Akter
Abstract:
This review paper identifies the core evidence of research on employee engagement , considering a stern challenge facing the financial sector nowadays. The study highlights the noteworthy knowledge gaps that will support human resource management practitioners to embed in the research towards sectoral context. Pertinent articles were selected through key search points and excerpt-related literatur…
▽ More
This review paper identifies the core evidence of research on employee engagement , considering a stern challenge facing the financial sector nowadays. The study highlights the noteworthy knowledge gaps that will support human resource management practitioners to embed in the research towards sectoral context. Pertinent articles were selected through key search points and excerpt-related literature. The key search points covered the topic related to different terms of engagement for example "employee engagement" OR "work engagement" OR "job engagement" OR "organization engagement" OR "staff engagement" OR "personnel engagement" which were steered in diverse context particularly financial sector. Through critically reviewing the literature for the last 11 years i.e., 2009-2019, we discovered 91 empirical studies in financial sector. From these studies, we found the overall concept of engagement and its different determinants (e.g., organizational factors, individual factors, job factors) as well as its various outcomes (e.g., employee outcomes, organizational outcomes). We also formulated a conceptual model to expand the body of knowledge in the area of employee engagement for a better understanding of its predictors and outcomes. Besides, limitations of the study and future recommendations are also contemplated.
△ Less
Submitted 11 June, 2021;
originally announced June 2021.
-
HomeGuard: A Smart System to Deal with the Emergency Response of Domestic Violence Victims
Authors:
Anik Islam,
Arifa Akter,
Bayzid Ashik Hossain
Abstract:
Domestic violence is a silent crisis in the developing and underdeveloped countries, though developed countries also remain drowned in the curse of it. In developed countries, victims can easily report and ask help on the contrary in developing and underdeveloped countries victims hardly report the crimes and when it's noticed by the authority it's become too late to save or support the victim. If…
▽ More
Domestic violence is a silent crisis in the developing and underdeveloped countries, though developed countries also remain drowned in the curse of it. In developed countries, victims can easily report and ask help on the contrary in developing and underdeveloped countries victims hardly report the crimes and when it's noticed by the authority it's become too late to save or support the victim. If this kind of problems can be identified at the very beginning of the event and proper actions can be taken, it'll not only help the victim but also reduce the domestic violence crimes. This paper proposed a smart system which can extract victim's situation and provide help according to it. Among of the developing and underdeveloped countries Bangladesh has been chosen though the rate of reporting of domestic violence is low, the extreme report collected by authorities is too high. Case studies collected by different NGO's relating to domestic violence have been studied and applied to extract possible condition for the victims.
△ Less
Submitted 25 March, 2018;
originally announced March 2018.
