Papers by Md. Sahidul Islam
Knowledge on Diabetes and Its Determinants among Type 2 Diabetic Subjects in a Low-Resource Setting: A Cross-Sectional Study in a Tertiary Care Hospital in Bangladesh, 2021
This is an open access journal, and articles are distributed under the terms of the Creative Comm... more This is an open access journal, and articles are distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 License, which allows others to remix, tweak, and build upon the work non-commercially, as long as appropriate credit is given and the new creations are licensed under the identical terms.
It is usual practice among researchers to use of graph of PC2 versus PC1 in order to visualize cl... more It is usual practice among researchers to use of graph of PC2 versus PC1 in order to visualize clusters in multivariate data. Recently some researchers suggest that ICA or ICA after PCA is better than PCA itself in this task. But no extensive comparison between these three techniques is available in literature. This article attempts to compare these on the basis of four sets of simulated data as well as three sets of real data. It is found that ICA after PCA is the best all the times, PCA alone fails in almost all the cases while ICA alone performs next to ICA after PCA. Theoretical justification behind this findings is also presented. So we recommend using of ICA after PCA to visualize clusters in multivariate data set. keywords: PCA, ICA, ICA on PCA, kurtosis. * M.Sc thesis student,
This thesis attempts to study ICA and compare it with PCA for detection of inherent structure, cl... more This thesis attempts to study ICA and compare it with PCA for detection of inherent structure, cluster analysis and outlier detection in multivariate data analysis. It presents the basic theory and application of ICA, and the recent work on
Conference Presentations by Md. Sahidul Islam
• Objective of this research • Multivariate statistical technique -Measure of Non-gaussianity W... more • Objective of this research • Multivariate statistical technique -Measure of Non-gaussianity Why we choose sub-gaussianity -Experimental setup Real and simulated dataset Result and discussion Principal Component analysis is the conventional technique for clustering approach in multivariate analysis. Independent component analysis is also a multivariate technique which has extensive use for clustering approach Independent component are deduced maximizing non gaussianity whereas PCA constructed maximizing variance.
Drafts by Md. Sahidul Islam
Sensory impaired individuals are at a disadvantage in accessing and processing electronic informa... more Sensory impaired individuals are at a disadvantage in accessing and processing electronic information. The first author (i.e. Thomas Hahn) is legally blind due to Albinism. In this experience report we describe challenges faced by the visually impaired and explain how remote access programs in combination with voice communication programs can be used to—at least partially—compensate for those disadvantages because they don't transmit magnification. This property is especially important to effectively train visually impaired individuals on new applications and interfaces remotely because it allows them to view exactly the same information simultaneously with their sighted trainers. Since the technical prerequisites for this information exchange, skill transfer, and knowledge acquisition approach have already been freely available for at least 7 years, but are still not widely used, this approach needs to be impressively demonstrated at conferences like this one to increase the odds that its participants will share this approach with those who could potentially benefit from it. This approach could make computer labs with expensive software not only accessible to the disabled, but instead, to everyone around the clock while saving money, which is still being spent to pay lab supervisors to keep the labs open for a few hours without losing—but instead—gaining functionality. Offering virtual remote office hours would benefit disabled and non-handicapped students and faculty alike. Providing remote access to lectures can make them available to a wider audience and thus could decrease costs for tuition. Obvious benefits of this approach for the mobility impaired and soon to be expected benefits for the hearing impaired are mentioned. Allowing faculty to remotely participate in oral exams increases choices for possible specializations. Making information more accessible to the disabled has obvious synergistic benefits for non-handicapped people alike as reflected by the importance of the concept of workforce diversification for overcoming unexpected future challenges and potential stumbling blocks. This approach makes it possible to magnify lecture presentations directly onto the screen of visually impaired students and could improve real-time interactions in the classroom. Remote access for everyone could reduce the perception of disabilities. Handicapped people could be considered early adopters because they are more in need of improvements since the present circumstances and limitations are much less acceptable to them. This article concludes by describing current bottlenecks to accessibility and information transfer, and ends with an overall optimistic outlook to the future. Figure 1: On the right laptop screen, the normal view of the programming interface of R Studio (see https://www.r-project.org/) is shown. The left laptop screen shows the alterations that are necessary for me to access and process electronic information. The setting displayed here closely simulates situations when sighted people train me to learn new interfaces or to follow instructions on how to use them.
Uploads
Papers by Md. Sahidul Islam
Conference Presentations by Md. Sahidul Islam
Drafts by Md. Sahidul Islam