Acne Scan - Application In Assessment Of Acne And Supports Treatment Road

As society develops, many aspects of life are concerned by people, including facial skincare, avoiding acne-related diseases. In this work, we will propose a complete solution for treating acne at home, including 4 processors. First, the anomaly detector uses image processing techniques by Multi-Threshold and Color Segmentation, depending on each color channel corresponding to each type of acne. The sensitivity of the detector is 89.4%. Second, the set of anomalies classifiers into 6 main categories, including 4 major acne types and 2 non-acne types. By applying the convolutional neural model, the accuracy, sensitivity, and F1 are 84.17%, 81.5%, and 82%, respectively. Third, the acne status assessment kit is based on the mGAGS method to classify the condition of a face as mild, moderate, severe, or very severe with an accuracy of 81.25%. Finally, the product recommender, which generalizes from the results of the previous processors with an accuracy of 70-90%. This is the premise that helps doctors as well as general users to evaluate the level of acne on a face effectively and save time.

Towards Intelligent Monitoring System in Wire Arc Additive Manufacturing: A Surface Anomaly Detector on a Small Dataset

Rapid developments in artificial intelligence and image processing have presented many new opportunities for defect detection in manufacturing processes. In this work, an intelligent image processing system has been developed to monitor inter-layer deposition quality during a Wire Arc Additive Manufacturing (WAAM) process. Information produced from this system is to be used in conjunction with other quality monitoring systems to verify the quality of fabricated components. It is tailored to identify the presence of defects relating to lack-of-fusion and voids immediately after the deposition of a given layer. The image processing system is built upon the YOLOv3 architecture and through moderate changes on anchor settings, achieves 53% precision on surface anomaly detection and 100% accuracy in identifying the fabricated components’ location, providing a prerequisite for high precision assessment of welding quality. The work presented in this paper presents an inter-layer vision-based defect monitoring system in WAAM and serves to highlight the feasibility of developing such intelligent computer vision systems for monitoring the WAAM process for defects.

Autoencoder-based anomaly detector for gear tooth bending fatigue cracks

The article reports on anomaly detection performance of data-driven models based on a few selected autoencoder topologies and compares them to the performance of a set of popular classical vibration-based condition indicators. The evaluation of these models employed data that consisted of baseline gearbox runs and the associated runs with seeded bending cracks in the root of the gear teeth for eight different gear pairings. The analyses showed that the data-driven models, trained on a subset of baseline data outperformed classical CIs as anomaly detectors.

Online Learning of Oil Leak Anomalies in Wind Turbines with Block-Based Binary Reservoir

The focus of this work is to design a deeply quantized anomaly detector of oil leaks that may happen at the junction between the wind turbine high-speed shaft and the external bracket of the power generator. We propose a block-based binary shallow echo state network (BBS-ESN) architecture belonging to the reservoir computing (RC) category and, as we believe, it also extends the extreme learning machines (ELM) domain. Furthermore, BBS-ESN performs binary block-based online training using fixed and minimal computational complexity to achieve low power consumption and deployability on an off-the-shelf micro-controller (MCU). This has been achieved through binarization of the images and 1-bit quantization of the network weights and activations. 3D rendering has been used to generate a novel publicly available dataset of photo-realistic images similar to those potentially acquired by image sensors on the field while monitoring the junction, without and with oil leaks. Extensive experimentation has been conducted using a STM32H743ZI2 MCU running at 480 MHz and the results achieved show an accurate identification of anomalies, with a reduced computational cost per image and memory occupancy. Based on the obtained results, we conclude that BBS-ESN is feasible on off-the-shelf 32 bit MCUs. Moreover, the solution is also scalable in the number of image cameras to be deployed and to achieve accurate and fast oil leak detections from different viewpoints.

An Ensembled Anomaly Detector for Wafer Fault Detection

The production process of a wafer in the semiconductor industry consists of several phases such as a diffusion and associated defectivity test, parametric test, electrical wafer sort test, assembly and associated defectivity tests, final test, and burn-in. Among these, the fault detection phase is critical to maintain the low number and the impact of anomalies that eventually result in a yield loss. The understanding and discovery of the causes of yield detractors is a complex procedure of root-cause analysis. Many parameters are tracked for fault detection, including pressure, voltage, power, or valve status. In the majority of the cases, a fault is due to a combination of two or more parameters, whose values apparently stay within the designed and checked control limits. In this work, we propose an ensembled anomaly detector which combines together univariate and multivariate analyses of the fault detection tracked parameters. The ensemble is based on three proposed and compared balancing strategies. The experimental phase is conducted on two real datasets that have been gathered in the semiconductor industry and made publicly available. The experimental validation, also conducted to compare our proposal with other traditional anomaly detection techniques, is promising in detecting anomalies retaining high recall with a low number of false alarms.

Understanding the Effect of Bias in Deep Anomaly Detection

Anomaly detection presents a unique challenge in machine learning, due to the scarcity of labeled anomaly data. Recent work attempts to mitigate such problems by augmenting training of deep anomaly detection models with additional labeled anomaly samples. However, the labeled data often does not align with the target distribution and introduces harmful bias to the trained model. In this paper, we aim to understand the effect of a biased anomaly set on anomaly detection. Concretely, we view anomaly detection as a supervised learning task where the objective is to optimize the recall at a given false positive rate. We formally study the relative scoring bias of an anomaly detector, defined as the difference in performance with respect to a baseline anomaly detector. We establish the first finite sample rates for estimating the relative scoring bias for deep anomaly detection, and empirically validate our theoretical results on both synthetic and real-world datasets. We also provide an extensive empirical study on how a biased training anomaly set affects the anomaly score function and therefore the detection performance on different anomaly classes. Our study demonstrates scenarios in which the biased anomaly set can be useful or problematic, and provides a solid benchmark for future research.

SALAD: An Exploration of Split Active Learning based Unsupervised Network Data Stream Anomaly Detection using Autoencoders

<div>Machine learning based intrusion detection systems monitor network data streams for cyber attacks. Challenges in this space include detection of unknown attacks, adaptation to changes in the data stream such as changes in underlying behaviour, the human cost of labeling data to retrain the machine learning model and the processing and memory constraints of a real-time data stream. Failure to manage the aforementioned factors could result in missed attacks, degraded detection performance, unnecessary expense or delayed detection times. This research evaluated autoencoders, a type of feed-forward neural network, as online anomaly detectors for network data streams. The autoencoder method was combined with an active learning strategy to further reduce labeling cost and speed up training and adaptation times, resulting in a proposed Split Active Learning Anomaly Detector (SALAD) method. The proposed method was evaluated with the NSL-KDD, KDD Cup 1999, and UNSW-NB15 data sets, using the scikit-multiflow framework. Results demonstrated that a novel Adaptive Anomaly Threshold method, combined with a split active learning strategy offered superior anomaly detection performance with a labeling budget of just 20%, significantly reducing the required human expertise to annotate the network data. Processing times of the autoencoder anomaly detector method were demonstrated to be significantly lower than traditional online learning methods, allowing for greatly improved responsiveness to attacks occurring in real time. Future research areas are applying unsupervised threshold methods, multi-label classification, sample annotation, and hybrid intrusion detection.</div>