Abstract
Predictive Maintenance applications are increasingly complex, with interactions between many components. Black -box models are popular approaches based on deep -learning techniques due to their predictive accuracy. This paper proposes a neural -symbolic architecture that uses an online rule -learning algorithm to explain when the black -box model predicts failures. The proposed system solves two problems in parallel: (i) anomaly detection and (ii) explanation of the anomaly. For the first problem, we use an unsupervised state-of-the-art autoencoder. For the second problem, we train a rule learning system that learns a mapping from the input features to the autoencoder's reconstruction error. Both systems run online and in parallel. The autoencoder signals an alarm for the examples with a reconstruction error that exceeds a threshold. The causes of the signal alarm are hard for humans to understand because they result from a non-linear combination of sensor data. The rule that triggers that example describes the relationship between the input features and the autoencoder's reconstruction error. The rule explains the failure signal by indicating which sensors contribute to the alarm and allowing the identification of the component involved in the failure. The system can present global explanations for the black box model and local explanations for why the black box model predicts a failure. We evaluate the proposed system in a real -world case study of Metro do Porto and provide explanations that illustrate its benefits.

Abstract
An algorithm for blind source separation (BSS) of the second heart sound (S2) into aortic and pulmonary components is proposed. It recovers aortic (A2) and pulmonary (P2) waveforms, as well as their relative delays, by solving an alternating optimization problem on the set of S2 sounds, without the use of auxiliary ECG or respiration phase measurement data. This unsupervised and data-driven approach assumes that the A2 and P2 components maintain the same waveform across heartbeats and that the relative delay between onset of the components varies according to respiration phase. The proposed approach is applied to synthetic heart sounds and to real-world heart sounds from 43 patients. It improves over two state-of-the-art BSS approaches by 10% normalized root mean-squared error in the reconstruction of aortic and pulmonary components using synthetic heart sounds, demonstrates robustness to noise, and recovery of splitting delays. The detection of pulmonary hypertension (PH) in a Brazilian population is demonstrated by training a classifier on three scalar features from the recovered A2 and P2 waveforms, and this yields an auROC of 0.76.

Abstract
In this study, different configurations based on linear fiber lasers were proposed and experimentally demonstrated to measure the concentration of liquid solutions. Samples of paracetamol liquid solutions with different concentrations, in the range from 52.61 to 201.33 g/kg, were used as a case-study. The optical gain was provided by a commercial bidirectional Erbium-Doped Fiber Amplifier (EDFA) and the linear cavity was obtained using two commercial Fiber Bragg Gratings (FBGs). The main difference of each configuration was the coupling ratio of the optical coupler used to extract the system signal. The sensing head corresponded to a Single-Mode Fiber (SMF) tip that worked as an intensity sensor. The results reveal that, despite the optical coupler used (50:50, 60:40, 70:30 or 80:20), all the configurations reached the laser condition, however, the concentration sensing was only possible using a laser drive current near to the threshold value. The configurations using a 70:30 and an 80:20 optical coupler allowed paracetamol concentration measurements with a higher sensitivity of (-3.00 +/- 0.24) pW/(g/kg) to be performed. In terms of resolution, the highest value obtained was 1.75 g/kg, when it was extracted at 20% of the output power to the linear cavity fiber laser configuration.

Abstract
Contrastive Analysis (CA) regards the problem of identifying patterns in images that allow distinguishing between a background (BG) dataset (i.e. healthy subjects) and a target (TG) dataset (i.e. unhealthy subjects). Recent works on this topic rely on variational autoencoders (VAE) or contrastive learning strategies to learn the patterns that separate TG samples from BG samples in a supervised manner. However, the dependency on target (unhealthy) samples can be challenging in medical scenarios due to their limited availability. Also, the blurred reconstructions of VAEs lack utility and interpretability. In this work, we redefine the CA task by employing a self-supervised contrastive encoder to learn a latent representation encoding only common patterns from input images, using samples exclusively from the BG dataset during training, and approximating the distribution of the target patterns by leveraging data augmentation techniques. Subsequently, we exploit state-of-the-art generative methods, i.e. diffusion models, conditioned on the learned latent representation to produce a realistic (healthy) version of the input image encoding solely the common patterns. Thorough validation on a facial image dataset and experiments across three brain MRI datasets demonstrate that conditioning the generative process of state-of-the-art generative methods with the latent representation from our self-supervised contrastive encoder yields improvements in the generated image quality and in the accuracy of image classification. The code is available at https://github.com/CristianoPatricio/unsupervised-contrastive-cond-diff.

Abstract
At a time when European countries try to cope with escalating energy prices while decarbonizing their economies, waste heat recovery and reuse arises as part of the solution for sustainable energy transitions. The lack of appropriate assessment tools has been pointed out as one of the main barriers to the wider deployment of waste heat recovery projects and as a reason why its potential remains largely untapped. The EMB3Rs platform emerges as an online, open-source, comprehensive and novel tool that provides an integrated assessment of different types of waste heat recovery solutions, (e.g. internal or external) and comprises several analysis dimensions (e.g. physical, geographical, technical, market, and business models). It has been developed together with stakeholders, and tested in a number of representative contexts, covering both industrial and heat network applications. This has demonstrated the enormous potential of the tool in dealing with complex simulations, while delivering accurate results within a significantly lower time-frame than traditional analysis. The EMB3Rs tool removes important barriers such as analysis costs, time and complexity for the user, and aims at supporting a wider investment in waste heat recovery and reuse by providing an integrated estimation of the costs and benefits of such projects. This paper describes the tool and illustrates how it can be applied to help unlock the potential of waste heat recovery across European countries.

Abstract
This paper presents an approach to enhancing sensitivity in optical sensors by integrating self-image theory and graphene oxide coating. The sensor is specifically engineered to quantitatively assess glucose concentrations in aqueous solutions that simulate the spectrum of glucose levels typically encountered in human saliva. Prior to sensor fabrication, the theoretical self-image points were rigorously validated using Multiphysics COMSOL 6.0 software. Subsequently, the sensor was fabricated to a length corresponding to the second self-image point (29.12 mm) and coated with an 80 mu m/mL graphene oxide film using the Layer-by-Layer technique. The sensor characterization in refractive index demonstrated a wavelength sensitivity of 200 +/- 6 nm/RIU. Comparative evaluations of uncoated and graphene oxide-coated sensors applied to measure glucose in solutions ranging from 25 to 200 mg/dL showed an eightfold sensitivity improvement with one bilayer of Polyethyleneimine/graphene. The final graphene oxide-based sensor exhibited a sensitivity of 10.403 +/- 0.004 pm/(mg/dL) and demonstrated stability with a low standard deviation of 0.46 pm/min and a maximum theoretical resolution of 1.90 mg/dL.