Search

In the rapidly evolving field of AI, Explainable Artificial Intelligence (XAI) has become paramount, particularly in Intelligent Environments applications. It offers clarity and understanding in complex decision-making processes, fostering trust and enabling rigorous scrutiny. The Shapley value, renowned for its accurate quantification of feature importance, has emerged as a prevalent standard in both academic research and practical application. Nevertheless, the Shapley value's reliance on the calculation of all possible coalitions poses a significant computational challenge, as it falls within the class of NP-hard problems. Consequently, approximation techniques are employed in most practical scenarios as a substitute for precise computations. The most common of those is the SHAP (SHapley Additive exPlanations) technique, which quantifies the influence exerted by a specific feature on decision outcomes of a specific Machine Learning model. However, the Shapley value's theoretical underpinnings focus on assessing and understanding feature impact on model evaluation metrics, rather than just alterations in the responses. This paper conducts a comparative analysis using controlled synthetic data with established ground truths. It juxtaposes the practical implementation of the SHAP approach with the theoretical model in two distinct scenarios: one using the F1-score and the other, the accuracy metric. These are two representative characteristic functions, capturing different aspects and whose appropriateness depends on the specific requirements and context of the task to be solved. We analyze how the three alternatives exhibit similarity and disparity in their manifestation of feature effects. We explore the parallels and differences between these approaches in reflecting feature effects. Ultimately, our research seeks to determine the conditions under which SHAP outcomes are more aligned with either the F1-score or the accuracy metric, thereby providing valuable insi

Regressive Remaining Useful Life Prediction and Survival Analysis are two lines of research with similar goals but different origins; one from engineering and the other from survival study in clinical research. Although the two research paths share a common objective of predicting the time to an event, researchers from each path typically do not compare their methods with methods from the other direction. Given the mentioned gap, we propose a framework to compare methods from the two lines of research using run-to-failure datasets. Then by utilizing the proposed framework, we compare six models incorporating three widely recognized degradation models along with two learning algorithms. The first dataset used in this study is C-MAPSS which includes simulation data from aircraft turbofan engines. The second dataset is real-world data from streamed condition monitoring of turbocharger devices installed on a fleet of Volvo trucks.

The new advances in Vehicular Systems and Technologies have resulted in a sheer increase in the number of connected vehicles. These connected vehicles use IoT technologies to communicate operational signals with the OEMs, such as the vehicle’s speed, torque, temperature, load, RPM, etc. These signals have provided an unprecedented opportunity to adaptively monitor the status of each piece of the vehicle’s equipment and discover any possible risk of failure before it happens. This emerging field of study is called predictive maintenance (also known as condition-based maintenance) and has recently received much attention. In this paper, we apply Integrated Gradients (IG), an XAI method until now primarily used on image data, on datasets containing tabular and time-series data in the domain of predictive maintenance of trucks’ turbochargers. We evaluate how the results of IG differ, in these new settings, for various types of models. In particular, we investigate how the change of baseline can affect the outcome. Experimental results verify that IG can be applied successfully to both sequenced and non-sequenced data. Contrary to the opinion common in the literature, the gradient baseline does not affect the results of IG significantly, especially on models such as RNN, LSTM, and GRU, where the data contains time series; the effect is more visible for models like MLP with non-sequenced data. To confirm these findings, and to understand them deeper, we have also applied IG to SVM models, which gave the results that the choice of gradient baseline has a significant impact on the performance of SVM. (c) IARIA, 2023

The discrepancy in the distribution of source and target domains is usually referred to as a domain shift. It is one of the reasons for the inferior performance of machine learning solutions at deployment. We illustrate that the domain shift issue is pertinent to the readings of the vehicles’ operational sensors. This is due to the fact that these measurements are collected over a period of time and are susceptible to various changes that happen in the meantime. Examples of these changes are usage pattern variations, aging of the vehicles, seasonal shifts, and driver changes. However, domain adversarial neural networks (DANN) have shown promising results to reduce the negative impact of the domain shift. The present study investigates domain adaptation (DA) in the predictive maintenance field by estimating the remaining useful life (RUL) of turbochargers. The devices are operating on a fleet of VOLVO trucks, and the information about their services is collected over four years between 2016 and 2019. The input features to the model are a set of bi-weekly collected measurements called logged vehicle data (LVD). The contributions of this paper are two-fold. First, we propose a new approach for detecting domain (covariate) shift using an autoencoder. Second, we adapt domain adversarial neural networks to the specific application of predicting turbocharger failures. Finally, we deploy a recurrent feature extraction layer in the DANN architecture to incorporate temporal aspect of the data. The experimental results demonstrate the superiority of the proposed method over the traditional approach.

Sensors are being used in many industrial applications for equipment health monitoring and anomaly detection. However, sometimes operation and maintenance of these sensors are costly. Thus companies are interested in reducing the number of required sensors as much as possible. The straightforward solution is to check the prediction power of sensors and eliminate those sensors with limited prediction capabilities. However, this is not an optimal solution because if we discard the identified sensors. As a result, their historical data also will not be utilized anymore. However, typically such historical data can help improve the remaining sensors' signal power, and abolishing them does not seem the right solution. Therefore, we propose the first data-driven approach based on tensor completion for re-utilizing data of removed sensors and the remaining sensors to create virtual sensors. We applied the proposed method on vibration sensors of high-speed separators, operating with five sensors. The producer company was interested in reducing the sensors to two. But with the aid of tensor completion-based virtual sensors, we show that we can safely keep only one sensor and use four virtual sensors that give almost equal detection power when we keep only two physical sensors., Project #280033, KK foundation, Strategic recruitment, Assistant Professor in information technology with focus on data mining.

Predicting energy (fuel and electric) consumption of hybrid vehicles is important on different levels: vehicle industry as a whole, individuals, and can also pave the way towards a more sustainable future. Despite its importance, providing accurate predictions is quite a challenging task. Many essential factors impacting energy consumption, including travel time, average speed, etc., needless to say, these features are not available beforehand. However, these factors are available in our data-set. To use these factors effectively, in this paper, we propose including them as different tasks in a multitask setting to help our main problem of energy consumption. The promise of this approach is that since these tasks are relevant, learning them together would provide a common feature space sharing information about all tasks. More importantly, this shared feature space would carry important information helping energy consumption in particular. In multitask learning, two important issues are task dominance and conflicting gradients of different tasks. Different studies have addressed these two separately. In this paper, we propose a method tackling these two problems simultaneously. We show experimentally the success of this method in comparison to state-of-the-art.

Extracting operation cycles from the historical reading of sensors is an essential step in IoT data analytics. For instance, we can exploit the obtained cycles for learning the normal states to feed into semi-supervised models or dictionaries for efficient real-time anomaly detection on the sensors. However, this is a difficult problem due to this fact that we may have different types of cycles, each of which with varying lengths. Current approaches are highly dependent on manual efforts by the aid of visualization and knowledge of domain experts, which is not feasible on a large scale. We propose a fully automated method called CycleFootprint that can: 1) identify the most relevant signal that has the most obvious recurring patterns among multiple signals; and 2) automatically find the cycles from the selected signal. The main idea behind CycleFootprint is mining footprints in the cycles. We assume that there should be a unique pattern in each cycle that shows up repeatedly in each cycle. By mining those footprints, we can identify cycles. We evaluate our method with existing labeled ground truth data of a real separator in marine application equipped with multiple health monitoring sensors. 86\% of cycles extracted by our method match fully or with at least 99\% overlap with true cycles, which sounds promising given its unsupervised and fully automated nature. © Springer Nature Switzerland AG 2020

The advancements of the telematics and connectivity solutions have provided new opportunities for the field of predictive maintenance. The number of sensors installed on a vehicle is increasing over time, and manufacturers are looking for new ways to improve the uptime of their fleet while at the same time reducing the costs related to unexpected breakdowns. The nature of the aggregated data from vehicles is sequential, and it is interesting to investigate existing methods for modeling partially observable state sequences to detect common patterns of failure. In this paper, we introduce a new approach for predicting turbocharger failures of Volvo trucks. The first step of the method deals with modeling a sequence of readouts from each vehicle using a Markov process. To do so, we identify the most informative signals and then employ spatial similarity clustering on the readouts. We interpret each cluster as a Markov state and further convert the history of a truck into a trajectory of states. This trajectory is then aligned with repairs information to form a standard sequence labeling problem. Finally, we train a hidden Markov model (HMM) classifier for assessing the health condition of the equipment. Empirical evaluations obtained on our realworld dataset of trucks suggest that the proposed method improves the AUC score of the final system up to 6% for predicting failures of a turbocharger.

In this study, we investigate the problem of detecting humans fall from video images. Many of the existing methods try to solve the problem by manually defining a set of hand-crafted features for detecting fall, which is not only a suboptimal approach but also cumbersome. On the contrary, the proposed method puts the burden of feature extraction on a pre-trained deep neural network. In this way, we can extract a comprehensive set of conceptual features automatically and efficiently. An important challenge of employing deep neural networks is the need for a large collection of training data. While the available labeled data for human fall detection is very limited, we propose three approaches based on transfer learning, and we trained them on two standard RGB and depth datasets for fall detection. The pre-trained models explored in this study are VGG16, Inception V3, and ResNet50. Support vector machine and logistic regression are used to classify the extracted features from videos into two classes of fall and normal daily activities. The experimental results obtained from the proposed approach suggest that the transfer learning tactic is able to compensate for the low training data issue. It is also shown that the proposed approach can efficiently extract important features from the sequences of video and boost the accuracy of the system on the task of human fall detection.

Monitoring the condition, detecting faults, and modeling the degradation of industrial equipment are important challenges in Prognostics and Health Management (PHM) field. Our solution to the challenge demonstrated a multi-stage approach for detecting faults in a group of identical industrial equipment, composed of four identical interconnected components, that have been deployed to the evolving environment with changes in operational and environmental conditions. In the first stage, a stacked ensemble of heterogeneous classifiers was applied to predict the state of each component of the equipment individually. In the second stage, a low pass filter was applied to smoothen the predictions cast by stacked ensembles, utilizing temporal information of the prediction sequence. © ESREL2020-PSAM15 Organizers. Published by Research Publishing, Singapore.