This paper presents an explainable ensemble learning framework using Bootstrap Aggregating to predict structural damage in masonry buildings during seismic events.It estimates the peak ground acceleration (PGA) leading to the damage control limit state (significant damage) based on structural parameters.The model achieves high accuracy (R 2 =0.9536,MAE=0.0057) and interpretability through SHAP, aligning with structural engineering principles.Compared to finite element analyses, it offers faster computations (milliseconds) and scalability, enabling rapid intervention planning after earthquakes.Developed under the MEDEA project (EU Grant n. 10101236), it supports disaster response and enhances seismic resilience.

Human pose estimation (HPE) models based on RGB images are widely used in applications such as surveillance, sports analytics, and healthcare. However, they often overlook privacy requirements mandated by data protection laws worldwide. One approach to ensure privacy is to use alternative data types such as LiDAR, WiFi signals, or depth images, but these can result in reduced accuracy compared to RGB-based models. A practical alternative is to apply obfuscation techniques to RGB images, which can help preserve privacy while retaining the benefits of high-resolution visual data. This paper investigates how HPE models perform under different types and intensities of image obfuscation. We focus on two common methods: Gaussian blur and pixelation, applied at multiple intensity levels. We evaluated three state-of-the-art HPE models, HRNet-W48, ViTPose Base, and RTMPose-l, in the COCO dataset, analyzing their robustness against obfuscated images. Key questions include how performance degrades with increasing obfuscation, which body parts are most affected, and the obfuscation thresholds where model fine-tuning becomes necessary. To complement this performance evaluation, we also show heatmaps to explain how obfuscation impacts the model focus on different body parts. Our study shows how to balance privacy and accuracy in HPE systems, offering practical guidance on using obfuscation techniques with existing RGB models to enhance privacy while maintaining performance without retraining.

Running injuries remain frequent, in particular among recreational athletes. Inappropriate foot-strike patterns are a recognized risk factor. We investigate step-wise classification of landing pattern (rear-foot vs. fore-foot) from wearable inertial measurements using time-domain features and machine learning. On a dataset of fourteen runners evaluated with leave-one-subject-out validation, Gradient Boosting achieved the best performance, with a median accuracy of 0.89 (range$0.73-1.00$) and a median weighted$\mathbf{F 1}$-score of 0.905. The practical value of distinguishing anterior from posterior strike lies in its combination with running intensity. This link enables actionable uses: risk analysis and injury prevention (for example, flagging persistent heel strikes at high intensity), gait optimization (promoting a recommended pattern for a given pace), and coach-supported feedback in training and rehabilitation. The results indicate that accurate, interpretable foot-strike recognition can support these applications in real-world running.

Multimedia AI now characterizes multiple concurrent streams at high resolutions, spanning video, audio, language, and multimodal fusion. Energy, bandwidth, and even water for cooling have become first-order constraints for real deployments. Training a single large NLP model has been estimated to emit up to 284 tons of$\text{CO}_{2}$, while infrastructure and model choices can shift emissions by orders of magnitude; at the same time, data centers consume huge volumes of water for cooling (hundreds of thousands). In this position paper, we argue that we cannot continue to choose models based on accuracy and throughput alone. We propose a Sustainability Card for Multimedia AI—Correct Outputs per kWh, Joules per Sample, Bytes per Output, and optional Water per 1,000 Outputs —and outline a simple measurement protocol.

For decades, RGB data has been central in human-pose estimation (HPE). The growing legislative pressure now limits their use during deployment, as RGB frames reveal sensitive personal information (SPI). Recent work often designs and trains new anonymization models, at a high computational cost. We ask instead whether adaptive filtering is sufficient. We introduce an adaptive obfuscation pipeline that converts RGB datasets into privacy-aware counterparts without training new networks. For each image, we detect the largest face, estimate a re-identification risk from its relative area, and raise blur or pixelation until an InsightFace recognizer fails to match the original embedding; that intensity is then applied to the whole image. Applied to MS-COCO, the pipeline generates two new sets, COCO-BLUR and COCO-PIX. On COCO-BLUR, three off-the-shelf HPE models retain more than 87% of their baseline AP. We release our code at this link (GitHub repository) offering a sustainable way to anonymize RGB data.

This position paper argues that human pose estimation (HPE) cannot be considered privacy-preserving or human-centric unless privacy is measured and evaluated. Although privacy concerns have become more visible in recent years, HPE systems are still assessed almost exclusively using accuracy metrics. Privacy is neither defined in measurable terms nor linked to regulatory requirements, and common deployment architectures introduce additional risks due to data transmission and storage. We highlight the limitations of current practices, including the continued reliance on RGB inputs and the lack of benchmarks that reflect legal and ethical constraints. We call for a shift in evaluation practices: privacy must become part of how HPE systems are designed, tested, and compared.

Poor posture is becoming more widespread due to the rising number of jobs that require workers to sit for extended hours. Maintaining proper leg positioning is essential for good overall posture and long-term health. However, current monitoring methods involve multiple sensors and cameras, leading to discomfort and privacy concerns. This paper presents a privacy-preserving recommendation system using Light Detection and Ranging (LiDAR) to monitor leg positions. The system captures the horizontal leg outline at knee height and extracts domain-specific features, while a stacked autoencoder reproduces the latent representation of the leg outline. The system comprises three modules, including one-class and multi-class support vector machines, to identify 15 leg positions and reject unrecognized ones. Data from 30 healthy volunteers during work activities trained the system. When tested on new participants, the system achieved an accuracy of over 98%. In addition to monitoring leg positions while respecting privacy and ergonomics, our system alerts workers about poor leg positions and generates intuitive dashboards with posture statistics that help safety engineers identify at-risk workers and body parts. This offers a promising solution for improving desk workers' posture and reducing long-term health issues, respecting privacy.

Every year, 2.5 million car crashes involve distracted drivers globally. It takes a few seconds for a car crash to happen after the driver has been distracted. Distracted driving thus poses a critical threat to road safety, needing innovative approaches for its detection and mitigation. This paper introduces a novel system to monitor in-car conversations and identify potential distractions from escalating arguments. The system analyzes Mel spectrograms generated from real-time audio signals containing in-car discussions by combining continuous voice recording and deep learning techniques. First, a denoiser employs a convolutional autoencoder to reduce car engine noise within the spectrograms. Then, a classifier uses convolutional and recurrent neural networks to determine whether the audio corresponds to a calm conversation or a quarrel based on the denoised spectrogram. The experimental results showed that the system achieved a 91.8% classification accuracy. This system addresses a previously unexplored dimension of cognitive distraction, offering valuable insights into strategies for reducing the risk of road accidents. Ongoing research is focused on accounting for other environmental noises, such as radio speakers, music, wind from open windows, and engine sounds from surrounding vehicles, which may influence classification accuracy. The system is also being extended to consider more than two occupants in the car.

The papers in this special issue focus on artificial intelligence (AI) for efficiency and sustainability in assembly/disassembly industrial processes. Assembly and disassembly lines are the backbone of modern manufacturing. These lines transform raw materials into finished products or take apart used products for reuse and recycling. Assembly lines are designed to optimize the process of putting together products, ensuring that each part is in the right place at the right time. Conversely, disassembly lines are designed to reverse this process, breaking down products into their components for reuse or recycling. Improving these lines is becoming more critical with the increasing complexity of products and the need for higher speed and greater efficiency.

Electroencephalography (EEG) is widely used for emotion recognition, but it typically requires cumbersome equipment used in lab settings. Modern EEG sensors are light and handy, and can be worn in daily life. However, these sensors have fewer electrodes than those used in labs which means less information available and the need for more sophisticated techniques for emotion recognition. This paper presents a system that combines Convolutional Neural Networks (CNNs) and Support Vector Machines (SVMs) to determine whether or not a person feels happiness based on a 14-channel EEG chart. The system first combines three pre-trained CNNs, namely, Inception V3, VGG-16, and ResNet-50 to extract features from an EEG chart represented as a binary matrix (binary-valued EEG chart). One module made up of three one-class SVMs is then used for each CNN to detect happiness (or non-happiness). The final class is that detected by the majority of the SVM modules. A dataset was collected involving 30 participants who watched 26 video clips as their EEG signals were being recorded by an EMOTIV EPOCXdevice. Participants self-rated their emotions using the 9-point Self-Assessment Manikin (SAM), and emojis. The system achieved an accuracy of 86.4%.

This paper presents an Android smartphone app to collect emotion-labeled physiological signals in the wild. The app controls a body sensor network that starts and stops sensors in random moments, without user knowing. Experiments involved recording the galvanic skin response (GSR), photoplethysmogram (PPG), 2-channel electromyogram (EMG), and 4-channel electroencephalogram (EEG) of 30 volunteers. AI models trained using the collected data determined a person’s valence and arousal levels with an accuracy of 92.8% and 94%, respectively. The app is free to download by contacting the authors.

Measuring the color difference between an image and a copy is crucial in the color industry, as many printing processes exist whose aim is to accurately reproduce colors. Due to various phenomena that are often unpredictable, these processes may print copies that are perceptually different from the original. Visual inspections are thus required in order to constantly control the color quality. These inspections are made by experts who observe a master image and a copy, typically at high resolution. This takes a long time for image acquisition, and generates high costs that could be reduced if these processes could be performed at lower resolution, possibly without human intervention, as people may perceive colors in different ways. This paper presents a neural network that assesses the color difference of multicolored images (photos) at different resolutions. The results showed that the level of color difference perceived by the network remains unchanged as the resolution decreases, and sometimes is even more evident.

Abstract Traffic and air pollution caused by the increasing number of cars have become important issues in nowadays cities. A possible solution is to employ recommender systems for efficient ridesharing among users. These systems, however, typically do not allow specifying ordered stops, thus preventing a large amount of possible users from exploiting ridesharing, e.g. parents leaving kids at school while going to work. Indeed, if a parent desired to share a ride, he/she would need to indicate the following constraint in the path: the stop at school should precede the stop at work. In this paper, we propose a ridesharing recommender, which allows each user to specify an ordered list of stops and suggests efficient ride matches. The ride-matching criterion is based on a dissimilarity between the driver’s path and the shared path, computed as the shortest path on a directed acyclic graph with ordering constraints between the stops defined in the single paths. The dissimilarity value is the detour requested to the driver to visit also the stops of the paths involved in the ride-share, respecting the visiting order of the stops within each path. Results are presented on a case study involving the city of Pisa.

Low back pain affects one in three workers in the world and is among the biggest causes of absence from work. Almost 75% of back injuries occur when lifting loads. In warehousing, agriculture, and construction, for example, workers are continuously handling loads manually. If incorrectly performed, these tasks put the workers at risk of back pain, injuries, and musculoskeletal disorders. Monitoring how the loads are lifted is key to quickly detecting which workers are showing dangerous behaviors, so that they can be (re)trained to perform the task safely, thereby reducing the risk of injury. This article presents a system based on artificial intelligence (AI) that exploits wearable sensors to assess the safety level of workers lifting loads. The system consists of a reflective safety jacket equipped with two barometric altimeters, a triaxial accelerometer, and a triaxial magnetometer. The sensors of the jacket continuously record these signals during the workday. The system then fuses the data from the two barometric altimeters in order to detect when the worker lifted loads. A neural classifier uses the signals recorded by the accelerometer and magnetometer to determine whether or not the task was performed safely. The system was tested on 30 workers and achieved an accuracy of 95.6%.

The refurbishment market is rich in opportunities-the global refurbished smartphones market alone will be $38.9 billion by 2025. Refurbishing a product involves disassembling it to test the key parts and replacing those that are defective or worn. This restores the product to like-new conditions, so that it can be put on the market again at a lower price. Making this process quick and efficient is crucial. This paper presents a novel formulation of parallel disassembly problem that maximizes the degree of parallelism, the level of ergonomics, and how the workers' workload is balanced, while minimizing the disassembly time and the number of times the product has to be rotated. The problem is solved using the Tensorial Memetic Algorithm (TeMA), a novel two-stage many-objective (MaO) algorithm, which encodes parallel disassembly plans by using third-order tensors. TeMA first splits the objectives into primary and secondary on the basis of a decision-maker's preferences, and then finds Pareto-optimal compromises (seeds) of the primary objectives. In the second stage, TeMA performs a fine-grained local search that explores the objective space regions around the seeds, to improve the secondary objectives. TeMA was tested on two real-world refurbishment processes involving a smartphone and a washing machine. The experiments showed that, on average, TeMA is statistically more accurate than various efficient MaO algorithms in the decision-maker's area of preference.

Printing is a widespread industrial process. Manufacturers of colored products are expected to maintain high levels of color quality to perfectly satisfy the customers' requirements. The rendering of colors is visually checked by experienced workers, who may though show different color sensitiveness, depending, e.g., on perceptual, cognitive and cultural aspects. This often results in products that fail to faithfully reproduce what the customer asked for, with negative consequences for companies, as well as huge financial losses. This paper describes a neural network-based system to objectively check how faithfully colors are reproduced by an industrial printing process. The system considers a master color, then compares it to a copy, and returns an objective degree of color fidelity of the copy to the master. The neural system was trained and tested in a real-world case study by using a huge quantity of color pairs taken from the L*a*b* color space. Highly accurate results were achieved. The strengths of the system are that it can measure the difference of colors in a way that is incredibly close to that perceived by the human eye, and the fact that it can do that canceling the color distortion phenomena that may occur in the human vision.

In a world where products get obsolescent ever more quickly, discarded devices produce million tons of electronic waste. Improving how end-of-life products are dismantled helps reduce this waste, as resources are conserved and fed back into the supply chain, thereby promoting reuse and recycling. This paper presents the Extremal MultiObjective Genetic Algorithm (EMOGA), a hybrid nature-inspired optimization technique for a multiobjective version of the disassembly line balancing problem. The aim is to minimize the number of workstations, and to maximize profit and disassembly depth, when dismounting products in disassembly lines. EMOGA is a Pareto-based genetic algorithm hybridized with a module based on extremal optimization, which uses a tailored mutation operator and a continuous relaxation-based seeding technique. The experiments involved the disassembly of a hammer drill and a microwave oven. Performance evaluation was carried out by comparing EMOGA to various efficient algorithms. The results showed that EMOGA is faster or gets closer to the Pareto front, or both, in all comparisons.

Every year 2.3 million people die worldwide due to occupational illnesses and accidents at work. By analyzing the workers' behavior when in the presence of risks managers could assign tasks to those workers who appear to be the most sensitive to risk being assigned and are thus more likely to exert more caution in the presence of that risk. This paper presents a novel multiobjective formulation of the personnel assignment problem, maximizing workers' sensitivity to risk, while minimizing cost and dislike for the task assigned. A worker's sensitivity to risk for a task is quantified by a new measure, carefulness, which stems from the worker's behavior and various human factors that affect the interaction with the risk. The problem is solved using a mixed evolutionary and multicriteria decision making methodology. An approximation of the Pareto front is first generated through the nondominated sorting genetic algorithm II. A hybrid decisional approach then exploits the technique for order of preference by similarity to ideal solution in order to select the Pareto-optimal solution that represents the nearest compromise to the decision-maker's preferences. These preferences are derived through a fuzzy version of the analytic hierarchy process. The proposed framework was tested in four real-world scenarios related to manufacturing companies. The results show a significant increase in overall carefulness and a strong decrease in the dislike for the task assigned, with a modest increase in cost. The framework thus improves the work climate and reduces the risk occurrence and/or the impact on the workers' health.

Worldwide, four people die every minute as a consequence of illnesses and accidents at work. This considerable number makes occupational safety an important research area aimed at obtaining safer and safer workplaces. This paper presents a semi-supervised learning-aided evolutionary approach to improve occupational safety by classifying workers depending on their own risk perception for the task assigned. More in detail, a semi-supervised learning phase is carried out to initialize a good population of a non-dominated sorting genetic algorithm (NSGA-II). Each chromosome of the population represents a pair of classifiers: one determines a worker's risk perception with respect to a task, the other determines the level of caution of the same worker for the same task. Learning from constraints reinforces the initial training performance. The best Pareto-optimal solution to the problem is selected by means of the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). The proposed framework was tested on real-world data gathered through a website purposely developed. Results showed a good performance of the obtained classifiers, thus validating the effectiveness of the proposed approach in supporting the decision-maker in critical job assignment problems, where risks are a serious threat to the workers' health.

We deal with the determination of the composition of bronze alloys measured through Laser-Induced Breakdown Spectroscopy (LIBS) analysis. The relation between LIBS spectra and bronze alloy composition, represented by means of the concentrations of constituting elements, is modeled by adopting an ensemble of learning machines, fed with different inputs. Then, the combiner computes the final response. The results obtained on the test set show that the ensemble model manages to determine the composition of alloy samples with mean squared error of about 6.53 10-2.

In spite of many improvements in industrial safety of the last decades, nowadays four people per minute die in the world for occupational illnesses and accidents at work. Besides equipping machines with the most advanced technologies, industrial safety has become more and more interested in human factors in recent years, since many accidents at work are proven to be blamed on dangerous behaviours of workers. Recruiting workers with proper risk perception and caution can increase how safely they will deal with the task assigned, thus reducing devastating events. This paper presents a many-objective optimization framework for personnel recruitment in safety-critical work environments. Four objectives are considered: cost and learning time (which are minimized), and risk perception and caution (which are maximized). A neural network-based module computes each candidate's risk perception and caution for every single task he/she applies for. Pareto optimal solutions are generated using the Multi-Objective Particle Swarm Optimizer based on hypervolume (MOPSOhv). The best personnel recruitment is selected by the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS). The effectiveness of the proposed framework was validated on two real-world recruitment processes involving 100 and 300 candidates, respectively.

Ride-sharing practice represents one of the possible answers to the traffic congestion problem in today's cities. In this scenario, recommenders aim to determine similarity among different paths with the aim of suggesting possible ride shares. In this paper, we propose a novel dissimilarity function between pairs of paths based on the construction of a shared path, which visits all points of the two paths by respecting the order of sequences within each of them. The shared path is computed as the shortest path on a directed acyclic graph with precedence constraints between the points of interest defined in the single paths. The dissimilarity function evaluates how much a user has to extend his/her path for covering the overall shared path. After computing the dissimilarity between any pair of paths, we execute a fuzzy relational clustering algorithm for determining groups of similar paths. Within these groups, the recommenders will choose users who can be invited to share rides. We show and discuss the results obtained by our approach on 45 paths.

Social networks have been recently employed as a source of information for event detection, with particular reference to road traffic congestion and car accidents. In this paper, we present a real-time monitoring system for traffic event detection from Twitter stream analysis. The system fetches tweets from Twitter according to several search criteria; processes tweets, by applying text mining techniques; and finally performs the classification of tweets. The aim is to assign the appropriate class label to each tweet, as related to a traffic event or not. The traffic detection system was employed for real-time monitoring of several areas of the Italian road network, allowing for detection of traffic events almost in real time, often before online traffic news web sites. We employed the support vector machine as a classification model, and we achieved an accuracy value of 95.75% by solving a binary classification problem (traffic versus nontraffic tweets). We were also able to discriminate if traffic is caused by an external event or not, by solving a multiclass classification problem and obtaining an accuracy value of 88.89%.

This paper presents a method, based on classification techniques, for automatic detection and diagnosis of defects of rolling element bearings. The experimental data set consists of vibration signals recorded by four accelerometers on a mechanical device including rolling element bearings: the signals were collected both with all faultless bearings and after substituting one faultless bearing with an artificially damaged one. Four defects and, for one of them, three severity levels are considered. Classification accuracy higher than 99% was achieved in all the experiments performed on the vibration signals represented in the frequency domain, thus proving the high sensitivity of our method to different types of defects and to different degrees of fault severity. The degree of robustness of our method to noise is also assessed by analyzing how the classification performance varies with the signal-to-noise ratio and using statistical classifiers and neural networks.

In this paper we describe a fuzzy rule-based classifier applied to forecasting of energy production in solar photovoltaic installations. After adapting the available numerical data to a dataset appropriate for classification, we propose a processing method to create an efficient rule base. The aim is to build an intelligent system able to forecast the class label of the energy production from a photovoltaic installation, given the values of some environmental parameters. Despite some already existing methods for forecasting problems, the main advantages of our approach are easier interpretability and versatility, as we deal with class labels. Moreover we propose a way to extract an ad hoc training dataset, in order to perform an effective training even when we deal with non optimal data (e.g., non-uniformly sampled data, missing samples, etc.). With a fuzzy forecasting system, in place of a traditional one, even the non-expert user of a photovoltaic system may be able to make decisions more easily. The results obtained show a correct classification percentage of almost 93%.

Situation-aware service recommendation for mobile devices is aimed at proactively pushing personalized suggestions to users, presenting them unseen or unknown services. A challenging area in the field is that of recommendation schemes emerging from users' collective behavior. When we consider a mobile user, for instance, the recommendation process can be based on social events that can arise from collective positioning information. In this scenario, we discuss a collaborative multi-agent scheme for event detection, in which fuzzy representations are employed to cope with the approximation typical of implicit and aggregated information. More specifically, the first level of information processing is managed by marking agents leaving marks in the environment which are associated with users' positioning. The accumulation of marks enables a fuzzy information granulation process, managed by event agents, in which relevant events can emerge. Finally, a fuzzy inference level, managed by situation agents, deduces user situations from the underlying events.

This paper presents a flexible approach to forecasting of energy production in solar photovoltaic (PV) installations, using time series analysis and neural networks. Our goal is to develop a one day-ahead forecasting model based on an artificial neural network with tapped delay lines. Despite some methods already exist for energy forecasting problems, the main novelty of our approach is the proposal of a tool for the technician of a PV installation to correctly configure the forecasting model according to the particular installation characteristics. The correct configuration takes into account the number of hidden neurons, the number of delay elements, and the training window width, i.e., the appropriate number of days, before the predicted day, employed for the training. The irradiation along with the sampling hour are used as input variables to predict the daily accumulated energy with a percentage error less than 5%.

In this work, an intelligent system for automatic detection of fault in PV fields is proposed. This system is based on a Takagi-Sugeno-Kahn Fuzzy Rule-Based System (TSK-FRBS), which provides an estimation of the instant power production of the PV field in normal functioning, i.e, when no faults occur. Then, the estimated power is compared with the real power and an alarm signal is generated if the difference between powers overcomes a threshold. The TSK-FRBS has been trained using data collected from a PV plant simulator, during normal functioning. Preliminary tests were carried out in a simulated framework, by reproducing both normal and fault conditions. Results show that the system can recognize more than 90% of fault conditions, even when noisy data are introduced.

In this paper, we propose a three-objective evolutionary algorithm to generate a set of Mamdani fuzzy rule-based systems (MFRBSs) with different tradeoffs between accuracy, rule base (RB) complexity and partition integrity. The RB, the linguistic partition granularities and the membership function (MF) parameters are concurrently learnt during the evolutionary process. In particular, the granularity learning is performed by exploiting the concept of virtual RB and an appropriate mapping strategy, and the MF parameter tuning is achieved by a piecewise linear transformation. The RB complexity is measured as the total number of conditions in the antecedents of the rules and the partition integrity is evaluated by using a purposely-defined index, based on the piecewise linear transformation. We use a chromosome composed of three parts, which codify, respectively, the RB, and, for each variable, the number of fuzzy sets and the parameters of the piecewise linear transformation of the membership functions. Results on two real-world regression problems are shown and discussed.

This paper first reviews the state-of-the-art of fuzzy rule-based classifiers (FRBCs), then it discusses how to implement an FRBC under the Pattern Recognition Toolbox (PRTools), the de-facto standard toolbox for classification in Matlab. Such an implementation, called frbc, allows for a straightforward comparison of frbc with other classifiers already available under the PRTools. Furthermore, frbc can easily be used and combined with any other general-purpose function already available in PRTools. In this way, e.g., it becomes really easy to perform many types of feature selection, based on the accuracy achieved by frbc on the subset of features at hand. Another useful feature is the capability to export each FRBC generated by frbc as a standard Fuzzy Inference System (FIS) structure used within the Matlab Fuzzy Logic Toolbox (FLT): this allows comparisons/validations, visual inspection of the rule base, etc. In the experimental part we first assess the correctness of the implementation, by reproducing results existing in the literature. Then we show some examples of usage of frbc, combined with existing PRTools functions.

In this paper, we compare traditional classifiers, such as Linear and Quadratic Discriminant Classifiers and neural networks, with a one-class classifier, namely, convex hull. With reference to rolling element bearing diagnosis, we show that convex hull outperforms traditional classifiers in the classification of faults and different levels of fault severity not known during the training phase.

Situation awareness is a promising approach to recommend to a mobile user the most suitable resources for a specific situation. However, determining the correct user situation is not a simple task since users have different habits that may affect the way in which the situations arise. Thus, an appropriate tuning aimed at adapting the situation recognizer to the specific user is desirable to make a resource recommender more effective. In this paper, we show how this objective can be achieved by collecting data during the interaction of the user with the mobile device and using this context history to personalize the resource recommender by a genetic algorithm. To describe our approach, we adopt a recently proposed resource recommender which exploits fuzzy linguistic variables to manage the inherent vagueness of some contextual parameters. Experimental results on a real business case show that the responsiveness and modeling capabilities of the recommender increase, thus validating the proposed approach.

Nowadays, a huge quantity of resources for mobile users are made available on the most important marketplaces. Further, handheld devices can accommodate plenty of these resources, such as applications, documents and web pages, locally. Thus, to search for resources suitable for specific circumstances often requires a considerable effort and rarely brings to a completely satisfactory result. A tool able to recommend suitable resources at the right time in each situation would be of great help for the mobile users and would make the use of the handheld devices less boring and more attractive. To this aim, new levels of granularity, together with some degree of self-awareness, are needed to assist mobile users in managing and using resources. In this paper, we propose an efficient situation-aware resource recommender (SARR), which helps mobile users to timely locate resources proactively. Situations are determined by a semantic reasoner that exploits domain knowledge expressed in terms of ontologies and semantic rules. This reasoner works in synergy with a fuzzy engine, which is in charge of handling the vagueness of some conditions in the semantic rules, computing a certainty degree for each inferred situation. These degrees are used to rank the situations and consequently to assign a priority to the resources associated with the specific situations. The application of SARR to two real business cases is also shown and discussed.

In this paper, we introduce a new index for evaluating the interpretability of Mamdani fuzzy rule-based systems (MFRBSs). The index takes both the rule base complexity and the data base integrity into account. We discuss the use of this index in the multi-objective evolutionary generation of MFRBSs with different trade-offs between accuracy and interpretability. The rule base and the membership function parameters of the MFRBSs are learnt concurrently by exploiting an appropriate chromosome coding and purposely-defined genetic operators. Results on a real-world regression problem are shown and discussed.

Today's mobile Internet service portals offer thousands of services and mobile devices can host plenty of applications, documents and web URLs. Hence, for average mobile users there is an increasing cognitive burden in finding the most appropriate service among the many available. On the other hand, methodologies such as bookmarks and resource tagging require a great arranging effort to handle increasing resources. To help mobile users in managing and using this personal information space, new levels of granularity should be introduced in the organization of services, together with some degree of self-awareness. This paper proposes a situation-aware service recommender that helps locating services proactively. In the recommender, a semantic layer determines one or more user current situations by using domain knowledge expressed in terms of ontology and semantic rules. A fuzzy inference layer manages the vagueness of some contextual condition of these rules and outputs an uncertainty degree for each situation. Based on this degree, the recommender proposes a set of specific resources.

This paper presents a method, based on classification techniques, for automatic detection and diagnosis of defects of rolling element bearings. We used vibration signals recorded by four accelerometers on a mechanical device including rolling element bearings: the signals were collected both with all faultless bearings and after substituting one faultless bearing with an artificially damaged one. We considered four defects and, for one of them, three severity levels. In all the experiments performed on the vibration signals represented in the frequency domain we achieved a classification accuracy higher than 99%, thus proving the high sensitivity of our method to different types of defects and to different degrees of fault severity. We also assessed the degree of robustness of our method to noise by analyzing how the classification performance varies on variation of the signal-to-noise ratio and using statistical classifiers and neural networks. We achieved very good levels of robustness.

This paper presents a method, based on classification techniques, for automatically detecting and diagnosing various types of defects which may occur on a rolling element bearing. In the experiments we have used vibration signals coming from a mechanical device including more than ten rolling element bearings monitored by means of four accelerometers: the signals have been collected both with all faultless bearings and substituting one faultless bearing with an artificially damaged one: four different defects have been taken into account. The proposed technique considers all the aspects of classification: feature selection, different base classifiers (two statistical classifiers, namely LDC and QDC, and MLP neural networks) and classifier fusion. Experiments, performed on the vibration signals represented in the frequency domain, have shown that the proposed classification method is highly sensitive to different types of defects and to different severity degrees of the defects.

In this paper, we propose the use of a multiobjective evolutionary approach to generate a set of linguistic fuzzy-rule-based systems with different tradeoffs between accuracy and interpretability in regression problems. Accuracy and interpretability are measured in terms of approximation error and rule base (RB) complexity, respectively. The proposed approach is based on concurrently learning RBs and parameters of the membership functions of the associated linguistic labels. To manage the size of the search space, we have integrated the linguistic two-tuple representation model, which allows the symbolic translation of a label by only considering one parameter, with an efficient modification of the well-known (2$+$2) Pareto Archived Evolution Strategy (PAES). We tested our approach on nine real-world datasets of different sizes and with different numbers of variables. Besides the (2$+$2)PAES, we have also used the well-known non-dominated sorting genetic algorithm (NSGA-II) and an accuracy-driven single-objective evolutionary algorithm (EA). We employed these optimization techniques both to concurrently learn rules and parameters and to learn only rules. We compared the different approaches by applying a nonparametric statistical test for pairwise comparisons, thus taking into consideration three representative points from the obtained Pareto fronts in the case of the multiobjective EAs. Finally, a data-complexity measure, which is typically used in pattern recognition to evaluate the data density in terms of average number of patterns per variable, has been introduced to characterize regression problems. Results confirm the effectiveness of our approach, particularly for (possibly high-dimensional) datasets with high values of the complexity metric.

We have developed a multi-classifier system for automatic classification of pulmonary nodules in lung CT (Computed Tomography) images. The system consists of a set of independent modules, each emulating a radiologist of a team, and a further module aimed at appropriately combining theradiologists' opinions. In the experiments we obtained a sensitivity of 95% against a specificity of 91.33%, adopting a combiner based on the decisio.n templates technique.

In this paper we propose a multi-objective genetic algorithm to generate Mamdani fuzzy rule-based systems with optimal trade-offs between complexity and accuracy. The main novelty of the algorithm is that both rule base and granularity of the uniform partitions defined on the input and output variables are learned concurrently. To this aim, we exploit a chromosome composed of two parts, which codify the numbers of fuzzy sets for each linguistic variable and the rule base, respectively. Rule bases defined on partitions with different granularity are handled by using an appropriate mapping strategy. The algorithm has been tested on a real word regression problem showing very promising results.

Multiobjective genetic fuzzy systems (MGFSs) have proved to be very effective in classification, regression and control tasks. However, large scale problems still present open and challenging research issues. Making identification of fuzzy rules faster can enlarge the range of applications of MGFSs. In this work we first analyze the time complexity for both the identification and the evaluation of Takagi-Sugeno fuzzy rule-based systems. Then we introduce a simple but effective idea for fast identification of consequent parameters, although in an approximated, suboptimal manner. In the experimental part we first test the correctness of the predicted asymptotical time complexity. Then we show the benefits through an example of multiobjective genetic learning of compact and accurate fuzzy systems, in which we saved 71.3% of time on a 7 input problem.

In this paper, we propose an approach to complexity reduction of Mamdani-type fuzzy rule-based systems (FRBSs) based on removing logical redundancies. We first generate an FRBS from data by applying a simplified version of the well-known Wang and Mendel method. Then, we represent the FRBS as a multi-valued logic relation. Finally, we apply MVSIS, a tool for circuit minimization and simulation, to minimize the relation and consequently to reduce complexity of the associated FRBS. Unlike similar previous approaches proposed in the literature, the use of MVSIS let us deal with nondeterminism, that is, let us manage rules with the same antecedent but different consequents. To allow nondeterminism guarantees to achieve a higher (or at least not worse) complexity reduction than the one achievable from removing nondeterminism as soon as it appears. We apply our approach to six popular benchmarks. Results show a considerable complexity reduction associated only sporadically with consistent accuracy degradation. Moreover, quite surprisingly, the complexity reduction often comes together with an improvement in the classification accuracy.

Context adaptation is certainly a promising approach in the development of fuzzy rule based systems (FRBSs). First, an initial rule base is extracted from heuristic knowledge of the application domain. Meanings of linguistic terms are defined so as to guarantee high interpretability of the FRBSs. Then, meanings are adapted to a specific context through the use of operators that, using a set of known input–output patterns, appropriately modify the corresponding fuzzy sets. The choice of the specific operators and their parameters is context based and optimized so as to obtain a good interpretability–accuracy trade-off. In this paper, we propose a set of operators that, starting from a given FRBS, adapt the FRBS to the specific context by adjusting the universes of the input and output variables, and modifying the core, the support and the shape of the fuzzy sets which compose the partitions of these universes. The operators are defined so as to preserve ordering of the linguistic terms, universality of rules, and interpretability of partitions. The choice of the parameters used in the operators is performed by a genetic optimization process aimed at maximizing the accuracy and preserving the interpretability of the FRBS. We finally describe the application of our context adaptation approach to two Mamdani fuzzy systems developed, respectively, for two different domains, namely, regression and data modeling. © 2008 Wiley Periodicals, Inc.

In this paper, we propose a novel population- based multi-objective evolutionary algorithm (MOEA) for binary classifier optimization. The two objectives considered in the proposed MOEA are the false positive rate (FPR) and the true positive rate (TPR), which are the two measures used in the ROC analysis to compare different classifiers. The main feature of our MOEA is that the population evolves based on the properties of the convex hulls defined in the FPR-TPR space. We discuss the application of our MOEA to determine a set of fuzzy rule-based classifiers with different trade-offs between FPR and TPR in lung nodule detection from CT scans. We show how the Pareto front approximation generated by our MOEA is better than the one generated by NSGA-II, one of the most known and used population-based MOEAs.

In the framework of context adaptation of fuzzy systems, a typical requirement of a contextualized system is to maintain the same interpretability as the original one. Here, we propose a novel index based on a fuzzy ordering relation to provide a measure of interpretability. Our index assesses ordering, distinguishability and coverage at the same time. We use the proposed index and the mean square error as goals of a multi-objective genetic algorithm aimed at generating contextualized Mamdani fuzzy systems with different trade-offs between the two goals. Results obtained on a synthetic data set are also discussed.

An approach to select the most suitable fuzzy rule-based binary classifier to a specific application is proposed. First, an evolutionary three-objective optimization algorithm is applied to generate an approximation of a Pareto front composed of fuzzy rule-based binary classifiers with different trade-offs between accuracy and complexity. Accuracy is measured in terms of sensitivity and specificity, whereas complexity is computed as sum of the conditions which compose the antecedents of the rules included in the classifiers. Thus, low values of complexity correspond to fuzzy systems characterized by a low number of rules and a low number of input variables actually used in each rule. This ensures a high comprehensibility of the classifiers. Then, the most suitable classifier is selected by using the ROC convex hull method. We discuss the application of the proposed approach to generate a classifier for discriminating lung nodules from non-nodules in a computer aided diagnosis (CAD) system. Results obtained on a real data set extracted from lung CT images are also discussed

Context adaptation can be achieved by adjusting an initial normalized fuzzy rule-based system through the use of operators that appropriately change the representation of the linguistic variables. The choice of the specific operators and their parameters should be context-based and optimized so as to obtain a good interpretability-accuracy tradeoff. In this paper we propose a set of context adaptation operators that, starting from a given fuzzy system, adjust some of its component!, such as fuzzy set support and core, membership function shape, etc. We use a genetic tuning process for choosing the operator parameters. We finally describe the application of the proposed operators to Mamdani fuzzy systems with reference to two real examples.

In this paper, we propose a fuzzy logic-based approach which exploits remotely sensed multispectral measurements of the reflected sunlight to estimate the concentration of optically active constituents of the sea water. The relation between the concentrations of interest and the subsurface reflectances is modeled by a set of fuzzy rules extracted automatically from the data through a two-step procedure. First, a compact initial rule base is generated by projecting onto the input variables the clusters produced by a fuzzy clustering algorithm. Then, a genetic algorithm is applied to optimize the rules. Appropriate constraints maintain the semantic properties of the initial model during the genetic evolution. Results of the application of the fuzzy model obtained from data simulated with an ocean color model over the channels of the Medium Resolution Imaging Spectrometer are shown and discussed.

One of the critical aspects of clustering algorithms is the correct identification of the dissimilarity measure used to drive the partitioning of the data set. The dissimilarity measure induces the cluster shape and therefore determines the success of clustering algorithms. As cluster shapes change from a data set to another, dissimilarity measures should be extracted from data. To this aim, we exploit some pairs of points with known dissimilarity value to teach a dissimilarity relation to a feed-forward neural network. Then, we use the neural dissimilarity measure to guide an unsupervised relational clustering algorithm. Experiments on synthetic data sets and on the Iris data set show that the relational clustering algorithm based on the neural dissimilarity outperforms some popular clustering algorithms (with possible partial supervision) based on spatial dissimilarity.

Most clustering algorithms partition a data set based on a dissimilarity relation expressed in terms of some distance function. When the nature of this relation is conceptual rather than metric, distance functions may fail to adequately model dissimilarity. For this reason, we propose to extract dissimilarity relations directly from the data. We exploit some pairs of patterns with known dissimilarity to build a TS fuzzy system, which models the dissimilarity relation between any pair of patterns. The resulting dissimilarity matrix is input to a new unsupervised fuzzy relational clustering algorithm, which partitions the data set based on the proximity of the vectors containing the dissimilarity values between a pattern and all the patterns in the data set. Experimental results to confirm the validity of our approach are shown and discussed.

In this paper, we present a system based on an Unsupervised Fuzzy Divisive Hierarchical Clustering (UFDHC) algorithm to determine a hierarchy of profiles of web site typical users from the web access log. These profiles can be extremely useful, for instance, to customize the web site, or to send personalized advertisements. After eliminating categories that have not been accessed by a significant percentage of users and removing the occasional users, the access log data are input to the UFDHC algorithm which clusters the users of the web site into a hierarchy of groups characterized by a set of common interests and represented by a prototype, which defines the profile of the group typical member. To show the effectiveness of our system, we describe how the profiles determined by the UFDHC algorithm from access log data collected along a period of 15 days allow classifying approximately 95% of the users defined by access log data collected during subsequent 60 days.

In this paper, we describe a fuzzy model for the estimation of sea water optically active constituents concentration in case II water. The model is extracted automatically from the data through a two step procedure. First, a fuzzy clustering algorithm is applied to identify a compact initial rule-based model. Then, the model is optimized by means of a genetic algorithm which tunes the rules so as to minimize the error between desired and predicted outputs. Appropriate constraints maintain the semantic properties of the initial model during the genetic evolution. The fuzzy model has been tailored to the multispectral data format of the MEdium Resolution Imaging Spectrometer (MERIS) on board the ESA-ENVISAT satellite launched in March 2002.

This paper describes a method for fuzzy classification and recognition of 2D shapes, such as handwritten characters, image contours, etc. A fuzzy model is derived for each considered shape from a fuzzy description of a set of instances of this shape. A fuzzy description of a shape instance, in its turn, exploits appropriate fuzzy partitions of the two dimensions of the shape. These fuzzy partitions allow us to identify, and automatically associate an importance degree with the relevant shape zones for classification and recognition purposes. Two significant applications of the method are described, namely, recognition of olfactory signals and recognition of isolated, handwritten characters. In the former case, results are shown concerning the recognition of three different types of waste waters, collected in three different dilutions. In the latter case, results are shown concerning the application of the method to a NIST database, containing the segmented handprinted characters of 500 writers.

Presents Beatrix, a system for off-line recognition of handwritten sentences. The system consists of four subsystems: the first pre-processes the input bitmap and splits each sentence into words; the second integrates neural and fuzzy logic techniques for individual character recognition; the third carries out a lexical and grammatical analysis of the recognised text; and the fourth uses the words recognised with a high confidence value to re-train the second subsystem on the specific handwriting. The analysis performed by the third subsystem produces hypotheses about words and sentences in order to correct errors made by the second subsystem. The fourth subsystem allows the system to self-adapt to the specific handwriting under examination; therefore, the system possesses the properties of both writer-independent and writer-dependent systems.

In this paper, we propose an evolutionary method for detecting the optimal number of clusters in a data set, and describe its application to classification of signals generated by olfactory sensors. The method is based on a new evolutionary search and optimization strategy. The strategy forces the formation and maintenance of subpopulations of solutions. Subpopulations co-evolve and converge towards different (sub-)optimal problem solutions. Only local chromosome interactions are allowed in order to avoid migration between subpopulations approximating different optimum points and to prevent the destruction of subpopulations. To this aim, specific selection and acceptance strategies have been defined. Experimental results obtained by applying the method to two test cases are also included.

In this paper we propose a new fuzzy linguistic method to recognize odorant samples. The method is applied to raw experimental data collected from a sensor array that comprises sixteen conducting polymer sensors with partially overlapping sensitivities. The sensors are exposed to odor samples and the percentage change in resistance is used for classification. The method describes the shape of each sensor response in terms of linguistic expressions derived from a fuzzy partition of the area occupied by the response. A purposely-defined weighted distance is used to compare the linguistic descriptions. Results on the application of the method to the classification of three chemicals in two different concentrations are presented.

In this paper we present SNIFFER, an electronic nose consisting of an array of conducting polymer sensors with partially overlapping sensitivities, and a pattern recognition system based on a new linguistic fuzzy classification method. The method describes the shape of the sensor responses in terms of linguistic expressions, which are derived from a fuzzy partition of the area occupied by each response. The sensors are exposed to odorants and the percentage change in resistance is used for classification purposes. Results of the application of SNIFFER to the recognition of different types of waste water produced both by the industrial activities of leather tanneries and by municipal sewers are presented.

Presents a fuzzy method for classification and recognition of separated handwritten characters. Linguistic expressions describing the individual characters are derived from a fuzzy model of a set of character samples. A small scale application of the method in which 26 lower-case cursive characters written by 30 different writers were analysed yielded 64% recognition rate. The method was also used to implement a character recogniser in a system for off-line recognition of cursive handwriting. In such a context, thanks to the use of a dictionary and a grammar parser, the recognition rate (at character level) rose to 96%.

This paper describes a method for constructing linguistic models from observed data. A linguistic model is derived from the reduction, based on clustering, of the number of fuzzy sets and rules which constitute a fuzzy model. This, in its turn, is built by applying a new method, called the local approximation method which determines a piecewise linear approximation of a set of samples of the system to be modelled. The approximation error due to linearisation can be chosen based on the degree of detail of the required model. In particular, if the final model is a linguistic one, the major requirements are readability and understandability, which normally correspond to reduced precision.

We consider fuzzy implication operators which are extensions of the two valued logic implication operator and are non decreasing with respect to their second argument. Firstly, we analyze some features of these operators with regard to fuzzy reasoning with one rule. Then, as regards approximate reasoning with multiple rules, we demonstrate that, for an inference process using Sup-T composition in the context of Compositional Rule of Inference (CRI), a necessary condition to infer a reasonable conclusion is that the minimum be used as an aggregation operator. Finally, when the minimum is used as an aggregation operator we provide a sufficient condition to obtain a reasonable inference result.

A distributed commit protocol suitable for multicomputer systems is described. A general programming environment which provides transactions as a programming tool is discussed. This environment is expected to be more dynamic than a database management system; in particular, it is not known how many and which processes will participate in a specific transaction. Therefore, a model of completely distributed transactions, without any hierarchical structure among the participant processes or any centralized locus of control, is proposed.>

In the event-action model of interactions between the debugging system and the program being debugged, an event will occur on the evaluation of a conditional defined in terms of the program activity if the evaluation yields the value true, and an action is an operation performed by the debugging system on the occurrence of an event. This paper presents a set of mechanisms for expressing conditionals at different levels of abstraction. At the lowest level, the authors have the simple conditionals, which can be expressed in terms of the values of the program entities and of the execution of the program statements. Simple conditionals can be grouped to form higher-level compound conditionals, which can be expressed in terms of the state and flow histories. The paper shows that the proposed abstraction mechanisms are powerful tools for monitoring program activity. They adequately support different debugging techniques, and offer the user a considerable degree of control over the debugging experiment.>