Search

Embracing Artificial Intelligence (AI) is becoming more common in a variety of areas, including healthcare, banking, and transportation, and it is based on substantial data analysis. However, utilizing data for AI raises a number of obstacles. This extensive article examines the challenges connected with using data for AI, including data quality, volume, privacy and security, bias and fairness, interpretability and ethical considerations, and the required technical knowledge. The investigation delves into each obstacle, providing insightful solutions for businesses and organizations to properly handle these complexities. Organizations may effectively harness AI's capabilities to make educated decisions by understanding and proactively tackling these difficulties, obtaining a competitive edge in the digital era. This review study, which provides a thorough examination of numerous solutions developed over the last decade to address data difficulties for AI, is expected to be a helpful resource for the scientific research community. It not only provides insights into current difficulties, but it also serves as a platform for creating novel ideas to alter our approaches to data strategies for AI. [ABSTRACT FROM AUTHOR]

Signature verification is done by using neural network to improve accuracy by reducing False Acceptance Rate (FAR) and False Rejection Rate(FRR). The two types of signature verification are online and offline signature verification. In this project offline signature verification is used in which signature is acquired using camera. Signature is checked by a specific resolution after the acquisition and preprocessing was done to remove the noise present in the image. Additional features are considered and extracted from an image and are trained using neural network to improve the accuracy. The same procedure is repeated for the test signature. The result of training and testing the signature leads to identify whether the signature is genuine or not with improved accuracy in comparison with conventional SVM technique. [ABSTRACT FROM AUTHOR]

The electrocardiogram shows the electrical activity of heart and is used by physicians to inspect the heart condition. Analysis of electrocardiogram becomes difficult if noise is embedded with signal during acquisition. In this paper, a de-noising technique for electrocardiogram signal done based on a combined method of high pass filtering with empirical mode decomposition (EMD) and wavelet de-noising. The major noise present in electrocardiogram signal is baseline wander and powerline interference. The aim of this proposed work is to remove both noises from the noise affected electrocardiogram signal. This work deals with the use of Butterworth high pass filter to de-noise the baseline wander followed by the empirical mode decomposition and wavelet denoising method to remove powerline interference noise completely. Finally, the signal to noise ratio is calculated for the proposed system. The high pass filtering with EMD de-noising is twice as better than de-noising of electrocardiogram using EMD technique only on an average. The Hybrid method consisting of high pass filtering, EMD with Wavelet which is about 4.5 times better on an average when compared to EMD technique only. [ABSTRACT FROM AUTHOR]

The electrocardiogram shows the electrical activity of heart and is used by physicians to inspect the heart condition. Analysis of electrocardiogram becomes difficult if noise is embedded with signal during acquisition. In this paper, a de-noising technique for electrocardiogram signal done based on a combined method of high pass filtering with empirical mode decomposition (EMD) and wavelet de-noising. The major noise present in electrocardiogram signal is baseline wander and powerline interference. The aim of this proposed work is to remove both noises from the noise affected electrocardiogram signal. This work deals with the use of Butterworth high pass filter to de-noise the baseline wander followed by the empirical mode decomposition and wavelet denoising method to remove powerline interference noise completely. Finally, the signal to noise ratio is calculated for the proposed system. The high pass filtering with EMD de-noising is twice as better than de-noising of electrocardiogram using EMD technique only on an average. The Hybrid method consisting of high pass filtering, EMD with Wavelet which is about 4.5 times better on an average when compared to EMD technique only. [ABSTRACT FROM AUTHOR]

The swift advancement of next-generation technologies for wireless communication has led to the cloud computing model becoming a more effective technique of data processing. In addition, cloud computing services are used in edge computing, often referred to as fog computing, to address new IoT challenges. Cloud computing is vulnerable to many infiltration flaws as a result of its Identity Server's shaky connection and dispersed communication features. Since distributed cloud computing systems are widely used, the great majority of research has concentrated on vulnerability scanning and defense. Nevertheless, there is still promise for the creation of fresh intrusion defense and detection methods. As a result, employing an edge computing-based topology of the network, this research proposed a novel hybrid optimal intelligence technique for the detection and authorization of anomalous dangerous behaviors in IoT networks. To discover anomalies in edge computing networks for detecting device behavior, a unique Evolving Neuro-Fuzzy Intelligence for Anomaly Detection (EFNI-AD) method is proposed in this research. Additionally, an Improved Cipher Crypto System (ICCS) is developed to ensure the reliability of data-passing edge devices. The edge device functions as a certification authority (CA) for the specified trusted domain. In this proposed ICCS paradigm, the edge node just verifies the certificate once to eliminate the overhead of edge devices, and once confidence has been established, all communication may be conducted utilizing regional credentials. The Lionized African Buffalo Optimization (LABO) also improves the assault detection system and trust authorization. The simulation findings indicate that the proposed methods outperform the current methods in terms of several performance criteria such as Packet delivery time, End-to-End delay, throughput, Routing overhead, Packet loss, Accuracy, processing time, Packet Delivery Ratio, and Recall. The proposed method provides lower end-to-end delay and a higher packet delivery ratio. [ABSTRACT FROM AUTHOR]