Breast tumor localization and segmentation using machine learning techniques: Overview of datasets, findings, and methods

Growing evidence suggests alterations in cognitive control processes in individuals with varying degrees of age-related hearing loss (ARHL); however, alterations in those with unaided mild ARHL are understudied. The current study examined two cognitive control processes, cognitive flexibility, and inhibition, in 21 older adults with unaided mild ARHL and 18 age- and education-matched normal hearing (NH) controls. All participants underwent comprehensive audiological and cognitive evaluations including Trail Making Test-B, Verbal Fluency, Stroop, and two Go/NoGo tasks. Group differences in cognitive flexibility and inhibition as well as associations between peripheral and central hearing ability and measures of cognitive flexibility and inhibition were investigated. Findings revealed that the ARHL group took significantly longer to complete the Stroop task and had higher error rates on NoGo trials on both Go/NoGo tasks relative to the NH controls. Additionally, poorer peripheral and central hearing were associated with poorer cognitive flexibility and inhibitory control. Our findings suggest slower and more inefficient inhibitory control in the mild ARHL group relative to the NH group and add to decades of research on the association between hearing and cognition.

https://www.mdpi.com/2308-3417/6/1/22/pdf

Cognitive Flexibility and Inhibition in Individuals with Age-Related Hearing Loss.

Knowledge maps are teaching tools that can promote deeply learning and avoid knowledge loss by helping students plan learning paths. Mining potential association rules of concepts from student exercise data was a common method to construct knowledge maps automatically. While manual conditions should be set to filter the association rules future to improve the accuracy of knowledge maps, which made the construction of the knowledge map can not automatic totally. So, the study proposed a knowledge map construction method that combined knowledge tracking and association rule mining expanding with interaction frequencies based on exercise data to achieve rules cleaning automatically. The method first predicted students’ knowledge mastery sequences by a deep knowledge tracking model and discovered clustering relations to represent potential structures between concepts by fuzzy cluster analysis. Meanwhile, the method investigated association rule mining expanding with interaction frequencies to discover association rules between concepts. Finally, the clustering relations were used to filter the mined association rules automatically. To verify the effectiveness of the method, we constructed a knowledge map based on 34,350 online exercise data of 117 students in a computer programming course. Experimental results proved that the map was valid. Our implementations are available at https://github.com/xxdmw/FPGF-master. 

Knowledge map construction based on association rule mining extending with interaction frequencies and knowledge tracking for rules cleaning

Due to the constant attack on network communications and mobile networks by hackers and the few security means of communication networks, a proposal to encrypt voice calls had to be developed. This proposal encrypts voice calls before they reach the phone and decrypts them when they are out of the phone at the receiver. The idea is to encrypt audio data without using servers or intermediary systems between the telephone device and the networks. CryptoBin encryption algorithm is used to encrypt and decrypt the audio data, which are characterized as a new and unknown algorithm, and characterized by speed and strength in encryption. This algorithm uses two secret keys to increase the strength of encryption and security.

Encryption Of Voice Calls Using CryptoBin Algorithm

In May 2022, it has received by WHO reports from non-endemic countries on cases of monkey pox disease. Monkey pox is a rare zoonotic disease caused by infection with the monkeypox virus that belongs to the genus orthopoxvirus and the family poxviridae, and also the variola virus. This study aims to classify patients who have contracted the monkey pox virus. We modeled an analysis of monkey pox disease and conducted comparisons utilizing a dataset from Kaggle consisting of a CSV file with records for 25,000 patients. The monkey pox dataset was analyzed using the correlation coefficient and the number of target variables.  Machine learning (ML) methods are used for classification by utilizing the K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Random Forest (RF), and Gradient Boosting (GB) algorithms. This study resulted in the highest classifier Gradient Boosting (GB) algorithm with an accuracy value of 71%. then the accuracy obtained by Support Vector Machine (SVM) is 69%, Random Forest (RF) accuracy is 68%, and finally K-Nearest Neighbor (KNN) obtains 63% accuracy. This ML method is expected to analyze monkey pox disease so that it helps the country and government, especially the health field in assessing, identifying, and being able to take appropriate action against monkey pox disease. 
 

/pdf/machine-learning-to-identify-monkey-pox-disease-14y3xkwz.pdf

Machine Learning to Identify Monkey Pox Disease

Current advances in parallel processing technology aims at providing unmatched degree of computational power in upcoming days. Parallel computation is an efficient form of information processing which exploits the concurrency of execution. This paper investigates the use of parallel programming, when applied on k Nearest Neighbors (kNN) algorithm which is intended for classification and prediction of the large dataset. Breast cancer dataset is used for classification and prediction which consists of two labels namely, malignant and benign. kNN is a non-parametric algorithm which makes use of similarity measure to classify the dataset into different categories. The similarity between the data points is computed by using Euclidean distance formula. Multiple threads are created for parallel processing and an appropriate kNN graph is constructed, which helps in easier implementation. Finally, execution speeds for sequential and parallel programs is recorded. The results are verified by using frameworks namely, Message Passing Interface (MPI) and Compute Unified Device Architecture (CUDA) highlighting that parallel execution takes less time when compared to sequential execution.

Parallel Implementation of kNN Algorithm for Breast Cancer Detection

Objective: The objective of this study was to investigate the association between hearing loss and cognitive functions in the elderly population. Methods: A cross-sectional study was carried out on 12 elderly individuals with hearing loss (elderly hearing impaired [EHI]) and 20 elderly individuals with normal hearing (elderly normal hearing [ENH]). The participants in the study were aged between 60 and 65 years. The audiological and cognitive tests were performed on both the groups. Results: The data compiled through descriptive statistics and multivariate analysis of variance revealed clinically significant differences between the EHI and ENH on different cognitive tests. This indicates that elderly population associated with hearing loss requires more time to process the tasks as compared to ENH individuals. Conclusion: There is a higher decline in cognitive functions in elderly individuals when associated with hearing loss compared to normal-hearing individuals. Since it is noticed as a comorbid condition, both cognition and hearing loss have to be assessed and treated in these individuals for better quality of life.

Comparison of cognitive functions in elderly population with and without hearing loss

The well-known multi-letter encryption cipher is Playfair cipher although a wide variety of techniques have been employed for encryption and decryption. The Playfair cipher shows a great advancement over other encryption methods. This cipher technique is used for encrypting and decrypting the pair of characters together. This paper proposes an idea to solve the traditional Playfair cipher through parallel algorithm in order to encrypt the given message. A key is used to encrypt the message. The idea is to reduce the time complexity of sequential Playfair cipher by parallelizing it and using the power of GPUs to the maximum. Instead of making one process do the encryption, a number of threads are used to encrypt letters simultaneously. This is achieved using CUDA.

Parallel Message Encryption Through Playfair Cipher Using CUDA

Many sorting algorithms have been developed over the years and the main aim is to reduce the time and space complexity for sorting the worst and average-case scenarios. Parallel computing greatly decreases the processing time and increases the processing speed. In this paper, we compare the results of a hybrid algorithm named CutShort algorithm using a parallel processing framework namely CUDA and MPI. We tested the proposed technique with random samples of large sample data. 30% speedup is achieved with parallel processing as compared to the sequential program.

Comparison of CutShort: A Hybrid Sorting Technique Using MPI and CUDA

Sorting algorithms have been a subject of research. Throughout the years various sorting algorithms have been implemented and their performance has been evaluated by comparing the space and time complexity. In this paper the super sort sorting algorithm with time complexity O(nlogn) has been implemented with MPI and CUDA. The intention is to compare the time taken by super sort algorithm when executed sequentially using C program and the time taken when implemented using CUDA and MPI.

Shwetha Rai

Papers

Parallel Implementation of kNN Algorithm for Breast Cancer Detection

Comparison of cognitive functions in elderly population with and without hearing loss

Parallel Message Encryption Through Playfair Cipher Using CUDA

Comparison of CutShort: A Hybrid Sorting Technique Using MPI and CUDA

Super Sort Algorithm Using MPI and CUDA