scholarly journals Robust Computing for Machine Learning-Based Systems

2020 ◽  
pp. 479-503
Author(s):  
Muhammad Abdullah Hanif ◽  
Faiq Khalid ◽  
Rachmad Vidya Wicaksana Putra ◽  
Mohammad Taghi Teimoori ◽  
Florian Kriebel ◽  
...  
Keyword(s):  
Machine Learning ◽  
Rapid Development ◽  
Autonomous Driving ◽  
High Accuracy ◽  
Security Threats ◽  
Security Vulnerabilities ◽  
Smart Healthcare ◽  
Super Computer ◽  
Temporal Redundancy ◽  
The Impact

AbstractThe drive for automation and constant monitoring has led to rapid development in the field of Machine Learning (ML). The high accuracy offered by the state-of-the-art ML algorithms like Deep Neural Networks (DNNs) has paved the way for these algorithms to being used even in the emerging safety-critical applications, e.g., autonomous driving and smart healthcare. However, these applications require assurance about the functionality of the underlying systems/algorithms. Therefore, the robustness of these ML algorithms to different reliability and security threats has to be thoroughly studied and mechanisms/methodologies have to be designed which result in increased inherent resilience of these ML algorithms. Since traditional reliability measures like spatial and temporal redundancy are costly, they may not be feasible for DNN-based ML systems which are already super computer and memory intensive. Hence, new robustness methods for ML systems are required. Towards this, in this chapter, we present our analyses illustrating the impact of different reliability and security vulnerabilities on the accuracy of DNNs. We also discuss techniques that can be employed to design ML algorithms such that they are inherently resilient to reliability and security threats. Towards the end, the chapter provides open research challenges and further research opportunities.

Security Threat Analyses and Attack Models for Approximate Computing Systems

2021 ◽  
Vol 26 (4) ◽  
pp. 1-31
Author(s):  
Pruthvy Yellu ◽  
Landon Buell ◽  
Miguel Mark ◽  
Michel A. Kinsy ◽  
Dongpeng Xu ◽  
...  
Keyword(s):  
Error Resilience ◽  
Security Threats ◽  
Approximate Computing ◽  
Security Threat ◽  
Security Vulnerabilities ◽  
Computing Systems ◽  
Quantitative Analyses ◽  
Resilience Mechanisms ◽  
The Impact ◽  
Attack Models

Approximate computing (AC) represents a paradigm shift from conventional precise processing to inexact computation but still satisfying the system requirement on accuracy. The rapid progress on the development of diverse AC techniques allows us to apply approximate computing to many computation-intensive applications. However, the utilization of AC techniques could bring in new unique security threats to computing systems. This work does a survey on existing circuit-, architecture-, and compiler-level approximate mechanisms/algorithms, with special emphasis on potential security vulnerabilities. Qualitative and quantitative analyses are performed to assess the impact of the new security threats on AC systems. Moreover, this work proposes four unique visionary attack models, which systematically cover the attacks that build covert channels, compensate approximation errors, terminate normal error resilience mechanisms, and propagate additional errors. To thwart those attacks, this work further offers the guideline of countermeasure designs. Several case studies are provided to illustrate the implementation of the suggested countermeasures.

Criticalities and Advantages of the Use of Artificial Intelligence in Research

2022 ◽  
pp. 161-175
Author(s):  
Jessica Camargo Molano ◽  
Jacopo Cavalaglio Camargo Molano
Keyword(s):  
Artificial Intelligence ◽  
Machine Learning ◽  
Public Schools ◽  
School System ◽  
Detection System ◽  
Rapid Development ◽  
Academic Research ◽  
The Face ◽  
The Impact ◽  
Machine Learning Models

In recent years, artficial intelligence, through the rapid development of machine learning and deep learning, has started to be used in different sectors, even in academic research. The objective of this study is a reflection on the possible errors that can occur when the analysis of human behavior and the development of academic research rely on artificial intelligence. To understand what errors artificial intelligence can make more easily, three cases have been analyzed: the use of the IMPACT system for the evaluation of school system in the District of Columbia Public Schools (DCPS) in Washington, the face detection system, and the “writing” of the first scientific text by artificial intelligence. In particular, this work takes into consideration the systematic errors due to the polarization of data with which the machine learning models are trained, the absence of feedback and the problem of minorities who cannot be represented through the use of big data.

scholarly journals Rationale Discovery and Explainable AI

2021 ◽  
Author(s):  
Cor Steging ◽  
Silja Renooij ◽  
Bart Verheij
Keyword(s):  
Machine Learning ◽  
Feature Detection ◽  
State Of The Art ◽  
High Accuracy ◽  
Learning Systems ◽  
Training Data ◽  
Relevant Feature ◽  
Explainable Ai ◽  
The Right ◽  
The Impact

The justification of an algorithm’s outcomes is important in many domains, and in particular in the law. However, previous research has shown that machine learning systems can make the right decisions for the wrong reasons: despite high accuracies, not all of the conditions that define the domain of the training data are learned. In this study, we investigate what the system does learn, using state-of-the-art explainable AI techniques. With the use of SHAP and LIME, we are able to show which features impact the decision making process and how the impact changes with different distributions of the training data. However, our results also show that even high accuracy and good relevant feature detection are no guarantee for a sound rationale. Hence these state-of-the-art explainable AI techniques cannot be used to fully expose unsound rationales, further advocating the need for a separate method for rationale evaluation.

scholarly journals The Impact of Psychopathology, Social Adversity and Stress-relevant DNAm on Prospective Risk for Post-traumatic Stress: A Machine Learning Approach

2020 ◽  
Author(s):  
Agaz H. Wani ◽  
Allison E. Aiello ◽  
Grace S. Kim ◽  
Fei Xue ◽  
Chantel L. Martin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Mean Square Error ◽  
Traumatic Stress ◽  
High Accuracy ◽  
Mean Square ◽  
Post Traumatic Stress ◽  
Social Adversity ◽  
Post Traumatic ◽  
Prospective Risk ◽  
The Impact

AbstractBackgroundA range of factors have been identified that contribute to greater incidence, severity, and prolonged course of post-traumatic stress disorder (PTSD), including: comorbid and/or prior psychopathology; social adversity such as low socioeconomic position, perceived discrimination, and isolation; and biological factors such as genomic variation at glucocorticoid receptor regulatory network (GRRN) genes. This complex etiology and clinical course make identification of people at higher risk of PTSD challenging. Here we leverage machine learning (ML) approaches to identify a core set of factors that may together predispose persons to PTSD.MethodsWe used multiple ML approaches to assess the relationship among DNA methylation (DNAm) at GRRN genes, prior psychopathology, social adversity, and prospective risk for PTS severity (PTSS).ResultsML models predicted prospective risk of PTSS with high accuracy. The Gradient Boost approach was the top-performing model with mean absolute error of 0.135, mean square error of 0.047, root mean square error of 0.217, and R2 of 95.29%. Prior PTSS ranked highest in predicting the prospective risk of PTSS, accounting for >88% of the prediction. The top ranked GRRN CpG site was cg05616442, in AKT1, and the top ranked social adversity feature was loneliness.ConclusionMultiple factors including prior PTSS, social adversity, and DNAm play a role in predicting prospective risk of PTSS. ML models identified factors accounting for increased PTSS risk with high accuracy, which may help to target risk factors that reduce the likelihood or course of PTSD, potentially pointing to approaches that can lead to early intervention.

scholarly journals Detect and Classify the Unpredictable Cyber-Attacks by using DNN Model

2021 ◽  
Vol 12 (6) ◽  
pp. 74-81
Author(s):  
G. Swarnalatha Et.al
Keyword(s):  
Machine Learning ◽  
Industrial Development ◽  
Rapid Development ◽  
Cyber Attacks ◽  
Machine Learning Algorithms ◽  
Machine Learning Techniques ◽  
Complete Analysis ◽  
Machine Learning Classification ◽  
Hyperparameter Selection ◽  
The Impact

Machine learning techniques are often used to develop IDS by detecting and deploying fast and automated network attacks to torpedoes and host standards. However, there are many problems, as severe attacks change all the time and occur at very high levels that require a lot of resolution. There are many malicious packages available for further investigation by the cybersecurity community. However, one completed study did not provide a complete analysis to apply different machine learning algorithms on different media packages. Because of the persistent methods of attack and the dynamic nature of malware, it is important to systematically update and approve malicious packages that are available to the public. This paper explores the DNN, a type of comprehensive learning model, promoting flexible and appropriate IDS for detecting and deploying expected and unpredictable online attacks. Sustainable industrial development and rapid development of attacks need evaluation for some data developed over the years using static and dynamic methods. This type of research can help determine the best algorithm to identify future attacks. Comparative data for some commonly available malware provides a comprehensive comparison of DNN experiences with other class machine learning classifications. The best network parameters and network topologies for DNN are selected using the KDDCup 99 package with this hyperparameter selection method. The DNN model, which works well on KDDCup 99, works on other data, such as the NSL-KDD memory test. Our DNN model teaches how to transfer IDS information functions from multicultural.Multidisciplinary representations in a variety of encryption. Complex tests have shown that DNN performs better than conventional machine learning classification. Finally, we present a large and hybrid DNN torrent structure called Scale-Hybrid-IDS-AlertNet, which can be used to effectively monitor the impact of network traffic and host-level events to warn directly about cyber-attacks.

Forecasting US movies box office performances in Turkey using machine learning algorithms

2020 ◽  
Vol 39 (5) ◽  
pp. 6579-6590
Author(s):  
Sandy Çağlıyor ◽  
Başar Öztayşi ◽  
Selime Sezgin
Keyword(s):  
Machine Learning ◽  
Global Economy ◽  
Learning Algorithms ◽  
Forecast Model ◽  
Machine Learning Algorithms ◽  
Gradient Boosting ◽  
High Stakes ◽  
Box Office ◽  
Industry Forecast ◽  
The Impact

The motion picture industry is one of the largest industries worldwide and has significant importance in the global economy. Considering the high stakes and high risks in the industry, forecast models and decision support systems are gaining importance. Several attempts have been made to estimate the theatrical performance of a movie before or at the early stages of its release. Nevertheless, these models are mostly used for predicting domestic performances and the industry still struggles to predict box office performances in overseas markets. In this study, the aim is to design a forecast model using different machine learning algorithms to estimate the theatrical success of US movies in Turkey. From various sources, a dataset of 1559 movies is constructed. Firstly, independent variables are grouped as pre-release, distributor type, and international distribution based on their characteristic. The number of attendances is discretized into three classes. Four popular machine learning algorithms, artificial neural networks, decision tree regression and gradient boosting tree and random forest are employed, and the impact of each group is observed by compared by the performance models. Then the number of target classes is increased into five and eight and results are compared with the previously developed models in the literature.

Research on behavior recognition based on feature fusion of automatic coder and recurrent neural network

2020 ◽  
Vol 39 (6) ◽  
pp. 8927-8935
Author(s):  
Bing Zheng ◽  
Dawei Yun ◽  
Yan Liang
Keyword(s):  
Neural Network ◽  
Recurrent Neural Network ◽  
Behavior Pattern ◽  
Rapid Development ◽  
Video Data ◽  
Support Vector ◽  
Behavior Recognition ◽  
Learning Methods ◽  
The Impact ◽  
Internet Of Things Technology

Under the impact of COVID-19, research on behavior recognition are highly needed. In this paper, we combine the algorithm of self-adaptive coder and recurrent neural network to realize the research of behavior pattern recognition. At present, most of the research of human behavior recognition is focused on the video data, which is based on the video number. At the same time, due to the complexity of video image data, it is easy to violate personal privacy. With the rapid development of Internet of things technology, it has attracted the attention of a large number of experts and scholars. Researchers have tried to use many machine learning methods, such as random forest, support vector machine and other shallow learning methods, which perform well in the laboratory environment, but there is still a long way to go from practical application. In this paper, a recursive neural network algorithm based on long and short term memory (LSTM) is proposed to realize the recognition of behavior patterns, so as to improve the accuracy of human activity behavior recognition.

The Scoping Review of Smart Healthcare Applications in Medical Health (Preprint)

2020 ◽  
Author(s):  
Cheng Hang Wu ◽  
Ching Ju Chiu ◽  
Yen Ju Liou ◽  
Chun Ying Lee ◽  
Susan C. Hu
Keyword(s):  
Systematic Reviews ◽  
Mobile Health ◽  
Scoping Review ◽  
Search Strategy ◽  
Rapid Development ◽  
Related Research ◽  
Geographic Access ◽  
Smart Healthcare ◽  
Research Findings ◽  
Healthcare Interventions

BACKGROUND There is still no consensus on research terms for smart healthcare worldwide. The study conducted by Lewis 10 years ago showed extending geographic access was the major health purpose of health-related information communication technology (ICT), but today's situation may be different because of the rapid development of smart healthcare. Objective: The main aim of this study is to classify recent smart healthcare interventions. Therefore, this scoping review was conducted as a feasible tool for exploring this domain and summarizing related research findings. OBJECTIVE The main aim of this study is to classify recent smart healthcare interventions. Therefore, this scoping review was conducted as a feasible tool for exploring this domain and summarizing related research findings. METHODS The scoping review relies on the analysis of previous reviews of smart healthcare interventions assessed for their effectiveness in the framework of a systematic review and/or meta-analysis. The search strategy was based on the identification of smart healthcare interventions reported as the proposed keywords. In the analysis, the reviews published from January 2015 to December 2019 were included. RESULTS The number of publications for smart healthcare's systematic reviews has continued to grow in the past five years. The search strategy yielded 210 systematic reviews and/or meta-analyses addressed to target groups of interest. 68.5% of these publications used mobile health as a keyword. According to the classification by Lewis, 37.62% of the literature was applied to extend geographic access. According to the classification by the Joint Commission of Taiwan (JCT), 48.84% of smart healthcare was applied in clinical areas, and 60% of it was applied in outpatient medical services. CONCLUSIONS Smart healthcare interventions are being widely used in clinical settings and for disease management. The research of mobile health has received the most attention among smart healthcare interventions. The main purpose of mobile health was used to extend geographic access to increase medical accessibility in clinical areas. CLINICALTRIAL none

Sign in / Sign up

Export Citation Format

Share Document