Intrusion refers to any malicious activity done in order to access confidential data. An intrusion detection system (IDS) detects these attacks and, on detection, it reports them to the administrator. It does so either by comparing the new activity with the past activities or by analyzing the network performance. This system forms a part of the vast security module and works with several other such sub-modules in order to make sure that these unwanted intrusions do not go unreported. The system that has been implemented in this paper is an anomaly-based Intrusion Detection System (IDS). The primary purpose of this implementation is to develop an efficient system in order to detect any external or internal unauthenticated activity. Several models have been experimented with in order to find one that suits the system the best and gives a good enough accuracy. The models that have been experimented with include Logistic Regressor, Random Forest Classifier, K Nearest Neighbor classifier, XGBoost Classifier, Gaussian Naive Bayes Classifier and a Multi-Layer Perceptron Classifier (MLP). Further, the accuracy of each of these models was calculated, and a comparative analysis was done between the performance of these models. The model that performed the best in this particular use case was the Random Forest Classifier giving an accuracy of 99.8% and a macro average F1-Score of 0.98.
In this paper, we propose a peak to average power ratio (PAPR) reduction scheme for generalised frequency-division multiplexing (GFDM) systems based on parallel concatenation of low-density parity-check codes (PC-LDPC) codes. The proposed scheme maps the PC-LDPC codewords onto subcarriers to construct a symbol of channel coded GFDM. Then, these sub symbols are combined with subcarriers to form symbols, and these symbols are applied to the PAPR expression of the GFDM system for calculation. The BER value is higher at 10dB for the RRC filter-based GFDM system than the RC filter. The RC filter configuration has less BER at 10dB in 0.2 roll-off factor value with 100 iterations and 200 iterations in soft decision algorithm. When the value of Complementary Cumulative Distribution Function (CCDF) =0.001, the PC-LDPC GFDM system reduces the PAPR by 4 to 4.5 dB compared to the uncoded GFDM signal. The coding gain of 0.5dB is observed in Raised cosine pulse shaping filter with PC-LDPC codes.
In deep learning solutions there has been a lot of ambiguity about how to make explainability inclusive of a machine learning pipeline. Recently, several deep learning techniques have been introduced to solve increasingly complicated problems with higher predictive capacity. However, this predictive power comes at the cost of high computational complexity and difficult to interpret. While these models often produce very accurate predictions, we need to be able to explain the path followed by such models for decision making. Deep learning models, in general, predict with no or very less interpretable explanations. This lack of explainability makes such models blackbox. Explainable Artificial Intelligence (XAI) aims at transforming this black box approach into a more interpretable one. In this paper, we apply the well known Grad-CAM technique for the explainability of tea-leaf classification problem. The proposed method classifies tea-leaf-bud combinations using pre-trained deep learning models. We add classification explainability in our tea-leaf dataset using the pre-trained model as an input to the Grad-CAM technique to produce class-specific heatmap. We analyzed the results and working of the classification models for their reliability and effectiveness.
Defects visualization in bare chips and dice are important for realization of high reliable aerospace systems. Various defects such as cracks, voids, delamination, chip outs, dis-coloration, deep scratches can be detrimental in long term reliability of the Hybrid Microcircuits (HMCs). In spite of various standards in the realization of dice as well as various inspection formulating methodologies still probability of defects induced during fabrication and packaging need to be critically examined. Further handling and assembly of the dice can also induce certain defects due to various stresses which will impact on product quality. This paper provides an overview of various visual defects observed in bare dice and further tests and analysis to know the impact on electronic fabrication process.
Internet of Things networks are becoming more popular for monitoring critical environments of various types, resulting in a large increase in the amount of data transmitted. Because of the large number of linked IoT devices, network and security protocols is a major concern. In the sphere of security, detection systems play a critical role: they are based on cutting-edge algorithms. They can recognize or forecast security attacks using techniques such as machine learning, allowing them to secure the underpinning system. We have depicted some of the Deep Learning based techniques and figured out the best technique called Long Short Term Memory (LSTM) with 87% of accuracy to build the Artificial Intelligence based Interpolation Technique for IoT Environment.
