Over the past seven decades since the advent of artificial intelligence (AI) technology, researchers have demonstrated and deployed systems incorporating AI in various domains. The absence of model explainability in critical systems such as medical AI and credit risk assessment among others has led to neglect of key ethical and professional principles which can cause considerable harm. With explainability methods, developers can check their models beyond mere performance and identify errors. This leads to increased efficiency in time and reduces development costs. The article summarizes that steering the traditional AI systems toward responsible AI engineering can address concerns raised in the deployment of AI systems and mitigate them by incorporating explainable AI methods. Finally, the article concludes with the societal benefits of the futuristic AI systems and the market shares for revenue generation possible through the deployment of trustworthy and ethical AI systems.
In this study, the authors explore the integration of speaker-independent automatic Amazigh speech recognition technology into interactive applications to extract data remotely from a distance database. Based on the combined interactive voice response (IVR) and automatic speech recognition (ASR) technologies, the authors built an interactive speech system to allow users to interact with the interactive system through voice commands. The hidden Markov models (HMMs), Gaussian mixture models (GMMs), and Mel frequency spectral coefficients (MFCCs) are used to develop a speech system based on the ten first Amazigh digits and six Amazigh words. The best-obtained performance is 89.64% by using 3 HMMs and 16 GMMs.
The primary challenge of intrusion detection systems (IDS) is to rapidly identify new attacks, learn from the adversary, and update the intrusion detection immediately. IDS operate in dynamic environments subjected to evolving data streams where data may come from different distributions. This is known as the problem of concept drift. Today's IDS though are equipped with deep learning algorithms most of the times fail to identify concept drift. This paper presents a technique to detect and adapt to concept drifts in streaming data with a large number of features often seen in IDS. The study modifies extreme gradient boosting (XGB) algorithm for adaptability of drifts and optimization for large datasets in IDS. The primary objective is to reduce the number of ‘false positives' and ‘false negatives' in the predictions. The method is tested on streaming data of smaller and larger sizes and compared against non-adaptive XGBoost and logistic regression.
Phony news or fake news spreads like a wildfire on social media causing loss to the society. Swift detection of fake news is a priority as it reduces harm to society. This paper developed a phony news detector for Reddit posts using popular machine learning techniques in conjunction with natural language processing techniques. Popular feature extraction algorithms like CountVectorizer (CV) and Term Frequency Inverse Document Frequency (TFIDF) were implemented. These features were fed to Multinomial Naive Bayes (MNB), Random Forest (RF), Support Vector Classifier (SVC), Logistic Regression (LR), AdaBoost, and XGBoost for classifying news as either genuine or phony. Finally, coefficient analysis was performed in order to interpret the best coefficients. The study revealed that the pipeline model of MNB and TFIDF achieved a best accuracy rate of 79.05% when compared to other pipeline models.
The scrambling of MOSFET below 22nm, 14nm, unwanted Short Channel Effects (SCE) like punch through, drain-induced barrier lowering (DIBL), along with huge leakage current are flowing through the device, which is not recognized for better performance. Multi-gate MOSFET generally measured as Fin-FET is the best substitute vital to stunned short channel effects. The work highlights results of the current-voltage electrical characteristics of the n-channel triple gate Fin-FET gatherings. The paper focuses on the study of geometry-based device design of Fin-FET by changing high k dielectrics materials from silicon SiO2 (3.9), Hafnium Oxide (HfO2), and metal gate work function ranging from 4.1eV to 4.5eV. The approach and simulation of 3Dimensional Fin-FET is carried to evaluate the better performance parameters of device for change in gate length by deploying different dielectrics materials. The effect on ratio of on current (ION) and off current (IOFF), threshold voltage (VTH), subthreshold slope (SS), and drain-induced barrier lowering (DIBL) is observed.
Emotions are described as strong feelings that are expressed by an individual in response to reactions to something or someone. Emotions are a very important aspect of day-to-day life interaction. Research shows that more than 90% of communication will happen non-verbally. This paper presents human emotion detection using a fuzzy relational model. The model consists of an image processing stage followed by an emotion recognition phase. The authors additionally made sub-categories in the most important expressions like happy and sad to discover the level of happiness and sadness in one face. Feature extraction along with multi-learning approach will help to test whether the person is truly happy or appearing to be happy. Experimental outcomes on the image dataset point out the accurate performance of the proposed technique. The experiment gives good accuracy results with the authors' own data set and robust with reference to some latest and leading edge.
With sensors percolating through everyday living, it may be toted that there is an enormous increase in the availability of real-time streaming and time series data. We also see an exponential increase in number of industry applications with sensors driven by IoT and connected with data sources that change over time. This time-series data presents many technical challenges, opportunities, and threats to industries. Thus, streaming analytics to model an unsupervised machine learning system for detecting unusual/anomalous behavior in real-time must be prominently addressed. In this paper, the authors propose a real-time abnormality detection model using a Facebook prophet that addresses issues related to the improper Facebook collection of data, further leading to faulty analysis and wrong results. The proposed unsupervised model detects abnormalities in the data captured through customer order by considering day and date as constraints. The proposed model is found to be even more efficient in RMSE score. The proposed model delivered enhanced performance compared to other traditional approaches.
Existing approaches to fruit detection experience difficulty in detecting small fruits with low overall detection accuracy. The reasons why many detectors are unable to handle small fruits better are that fruit data sets are small, and they are not enough to train previous models of YOLO. Further, these models used in fruit detection are initialized by a pre-trained model and then fine-tuned on fruit data sets. The pre-trained model was trained on the ImageNet data set whose objects have a bigger scale than that of the fruits in the fruit pictures. Fruit detection being a fundamental task for automatic yield estimation, the goal is to detect all the fruits in images. YOLO-V3 uses multi-scale prediction to detect the final target, and its network structure is more complex. Thus, in this work, YOLO-V3 is used to predict bounding boxes on different scales and to make multi-scale prediction, thereby making YOLO-V3 more effective for detecting small targets. The feature pyramid mechanism integrates multi-scale feature information to improve the detection accuracy.
This research work specifically focusses on the development of a predictive model for movie review data using support vector machine (SVM) classifier with its improvisations using different kernel functions upon sentiment score estimation. The predictive model development proceeds with user level data input with the data processing with the data stream for analysis. Then formal calculation of TF-IDF evaluation has been made upon data clustering using simple k-means algorithm. Once the labeled data has been sorted out, then the SVM with kernel functions corresponding to linear, sigmoid, rbf, and polynomial have been applied over the clustered data with specific parameter setting for each type of library functions. Performance of each of the kernels has been measured using precision, recall, and F-score values for each of the specified kernel, and from the analysis, it has been found that sentiment analysis using SVM linear kernel with sentiment score analysis has been found to provide an improved accuracy of about 91.18%.
Automatic text summarization has recently become a key instrument for reducing the huge quantity of textual data. In this paper, the authors propose a quantum-inspired genetic algorithm (QGA) for extractive single-document summarization. The QGA is used inside a totally automated system as an optimizer to search for the best combination of sentences to be put in the final summary. The presented approach is compared with 11 reference methods including supervised and unsupervised summarization techniques. They have evaluated the performances of the proposed approach on the DUC 2001 and DUC 2002 datasets using the ROUGE-1 and ROUGE-2 evaluation metrics. The obtained results show that the proposal can compete with other state-of-the-art methods. It is ranked first out of 12, outperforming all other algorithms.
This article performs the sclera segmentation task by proposing a new hybrid symbolic fuzzy c-means (HSFCM) clustering method. Practically, though the data point exhibits some sort of similarity, unfortunately they are not undistinguishable and exhibit some sort of dissimilarity. Thus, to capture these disparities, the proposed work uses symbolic interval valued representation method. Further, to handle uncertainty and imprecision, the paper has proposed to use symbolic fuzzy clustering methods. To assess the performance of the proposed method, extensive experimentation is conducted on SSRBC2016 dataset. The proposed clustering method is compared with existing FCM, KFCM, RSKFCM method in terms of cluster validity indices and accuracy. The obtained outcomes demonstrated that the proposed method performed better compared to the contemporary methods.
In this paper, the authors present a concatenation of Hartmann and Rudolph (HR) partially exploited and a decoder based on hash techniques and syndrome calculation to decode linear block codes. This work consists firstly to use the HR with a reduced number of codewords of the dual code then the HWDec which exploits the output of the HR partially exploited. Researchers have applied the proposed decoder to decode some Bose, Chaudhuri, and Hocquenghem (BCH) and quadratic residue (QR) codes. The simulation and comparison results show that the proposed decoder guarantees very good performances, compared to several competitors, with a much-reduced number of codewords of the dual code. For example, for the BCH(31, 16, 7) code, the good results found are based only on 3.66% of the all codewords of the dual code space, for the same code the reduction rate of the run time varies between 78% and 90% comparing to the use of Hartmann and Rudolph alone. This shows the efficiency, the rapidity, and the reduction of the memory space necessary for the proposed concatenation.
This paper describes the algorithm used for personal identification based on features extracted from the palmprint. The local Gabor XOR (LGXP) features is built using Gabor filter with orientation. Initially, the palm print images are preprocessed using median filter. The algorithm is then modified, where features are extracted with different orientations of the Gabor filter called the multiple orientation LGXP (MOLGXP) features. The PCA feature is extracted and fused with MOLGXP and PCA using sum rule.
The relevance and recompense offered by cloud services make it obligatory for researchers to address the issues faced by this paradigm. The two major issues faced by this fast growing and most rising technology of the last one and a half decades is trust and security. Although the literature has witnessed and practiced a lot of models that can address these issues, despite the efforts, a sharp increase can be seen in data leakage incidences and its leads to hamper trust of the consumers in the cloud services and service providers. In order to remove the obstacles that are hampering the growth of the technology. The authors have tried to work on the areas which are not yet explored to solve the trust problem which is an indirect outcome of security lapse instances. The idea is to utilize the power of certifications or implications of the following standards or working according to the guidelines mentioned. The objective of this research article is to analyze the service provider on these aspects and generate its trustworthiness.
Rust detection is necessary for proper working and maintenance of machines for security purposes. Images are one of the suggested platforms for rust detection in which rust can be detected even though the human can't reach to the area. However, there are a lack of online databases available that can provide a sizable dataset to identify the most suitable model that can be used further. This paper provides a data augmentation technique by using Perlin noise, and further, the generated images are tested on standard features (i.e., statistical values, entropy, along with SIFT and SURF methods). The two most generalized classifiers, naïve Bayes and support vector machine, are identified and tested to obtain the performance of classification of rusty and non-rusty images. The support vector machine provides better classification accuracy, which also suggests that that the combined features of statistics, SIFT, and SURF are able to differentiate the images. Hence, it can be further used to detect the rust in different parts of machines.
With rapid technological advancements and enhanced network growth, security contends to play a crucial role. A powerful network security tends to point out diverse mixture of threats and intimidations and blocks them from creeping and getting circulated into the network to preserve the reliability, confidentiality, integrity, and accessibility of computer networks by annihilating illegitimate admittance and corruption of critical information. Secure hash algorithms (SHA) are cryptographic hash functions used to produce a hash value of fixed output bit sizes. In this paper, an algorithm is proposed to strengthen the cryptographic systems by using reversible logic to generate higher and variable hash values, making it difficult to trace the keys. The proposed scheme is simulated and verified using FPGA Virtex ML505 board, the analysis of power and time of which is carried out using Genus tool, proving it to be efficient in terms of power, gate usage, garbage, and quantum cost.
Automatic recognition of facial expressions and modeling of human expressions are very essential in the field of affective computing. The authors have introduced a novel geometric and texture-based method to extract the shapio-geometric features from an image computed by landmarking the geometric locations of facial components using the active appearance model (AAM). Expression-specific analysis of facial landmark points is carried out to select a set of landmark points for each expression to identify features for each specific expression. The shape information matrix (SIM) is constructed the set salient landmark points assign to an expression. Finally, the histogram-oriented gradients (HoG) of SIM are computed which is used for classification with multi-layer perceptron (MLP). The proposed method is tested and validated on four well-known benchmark databases, which are CK+, JAFFE, MMI, and MUG. The proposed system achieved 98.5%, 97.6%, 96.4%, and 97.0% accuracy in CK+, JAFFE, MMI, and MUG database, respectively.
Automatic anatomical therapeutic chemical (ATC) classification predicts an unknown compound's therapeutic and chemical characteristics. Predicting the organs/systems an unidentified compound will act on has the potential of expediting drug development and research. That a given compound can affect multiple organs/systems makes automatic ATC classification a complex problem. In this paper, the authors experimentally develop a multi-label ensemble for ATC prediction. The proposed approach extracts a 1D feature vector based on a compound's chemical-chemical interaction and its structural and fingerprint similarities to other compounds, as defined by the ATC coding system. This 1D vector is reshaped into 2D matrices and fed into seven pre-trained convolutional neural networks (CNN). A bidirectional long short-term memory network (BiLSTM) is trained on the 1D vector. Features extracted from both deep learners are then trained on multi-label classifiers, with results fused. The best system proposed here is shown to outperform other methods reported in the literature.
Archaeological departments throughout the world have undertaken massive digitization projects to digitize their historical document corpus. In order to provide worldwide visibility to these historical documents residing in the digital libraries, a character recognition system is an inevitable tool. Automatic character recognition is a challenging problem as it needs a cautious blend of enhancement, segmentation, feature extraction, and classification techniques. This work presents a novel holistic character recognition system for the digitized Estampages of Historical Handwritten Kannada Stone Inscriptions (EHHKSI) belonging to 11th century. First, the EHHKSI images are enhanced using Retinex and Morphological operations to remove the degradations. Second, the images are segmented into characters by connected component labeling. Third, LBP features are extracted from these characters. Finally, decision tree is used to learn these features and classify the characters into appropriate classes. The LBP features improved the performance of the system significantly.
