Data-stream driven Fuzzy-granular approaches for system maintenance

Intelligent systems are currently inherent to the society, supporting a synergistic human-machine collaboration. Beyond economical and climate factors, energy consumption is strongly affected by the performance of computing systems. The quality of software functioning may invalidate any improvement attempt. In addition, data-driven machine learning algorithms are the basis for human-centered applications, being their interpretability one of the most important features of computational systems. Software maintenance is a critical discipline to support automatic and life-long system operation. As most software registers its inner events by means of logs, log analysis is an approach to keep system operation. Logs are characterized as Big data assembled in large-flow streams, being unstructured, heterogeneous, imprecise, and uncertain. This thesis addresses fuzzy and neuro-granular methods to provide maintenance solutions applied to anomaly detection (AD) and log parsing (LP), dealing with data uncertainty, identifying ideal time periods for detailed software analyses. LP provides deeper semantics interpretation of the anomalous occurrences. The solutions evolve over time and are general-purpose, being highly applicable, scalable, and maintainable. Granular classification models, namely, Fuzzy set-Based evolving Model (FBeM), evolving Granular Neural Network (eGNN), and evolving Gaussian Fuzzy Classifier (eGFC), are compared considering the AD problem. The evolving Log Parsing (eLP) method is proposed to approach the automatic parsing applied to system logs. All the methods perform recursive mechanisms to create, update, merge, and delete information granules according with the data behavior. For the first time in the evolving intelligent systems literature, the proposed method, eLP, is able to process streams of words and sentences. Essentially, regarding to AD accuracy, FBeM achieved (85.64+-3.69)%; eGNN reached (96.17+-0.78)%; eGFC obtained (92.48+-1.21)%; and eLP reached (96.05+-1.04)%. Besides being competitive, eLP particularly generates a log grammar, and presents a higher level of model interpretability.

Congenital Nystagmus eye movements analysis and oculomotor system models evaluation

The term Congenital Nystagmus (Early Onset Nystagmus or Infantile Nystagmus Syndrome) refers to a pathology characterised by an involuntary movement of the eyes, which often seriously reduces a subject’s vision. Congenital Nystagmus (CN) is a specific kind of nystagmus within the wider classification of infantile nystagmus, which can be best recognized and classified by means of a combination of clinical investigations and motility analysis; in some cases, eye movement recording and analysis are indispensable for diagnosis. However, interpretation of eye movement recordings still lacks of complete reliability; hence new analysis techniques and precise identification of concise parameters directly related to visual acuity are necessary to further support physicians’ decisions. To this aim, an index computed from eye movement recordings and related to the visual acuity of a subject is proposed in this thesis. This estimator is based on two parameters: the time spent by a subject effectively viewing a target (foveation time - Tf) and the standard deviation of eye position (SDp). Moreover, since previous studies have shown that visual acuity largely depends on SDp, a data collection pilot study was also conducted with the purpose of specifically identifying eventual slow rhythmic component in the eye position and to characterise in more detail the SDp. The results are presented in this thesis. In addition, some oculomotor system models are reviewed and a new approach to those models, i.e. the recovery of periodic orbits of the oculomotor system in patients with CN, is tested on real patients data. In conclusion, the results obtained within this research consent to completely and reliably characterise the slow rhythmic component sometimes present in eye position recordings of CN subjects and to better classify the different kinds of CN waveforms. Those findings can successfully support the clinicians in therapy planning and treatment outcome evaluation.

Assessment of Daily Life Mobility Levels using Wearable Inertial Sensors and Minimun Measured Input Models

Tracking activities during daily life and assessing movement parameters is essential for complementing the information gathered in confined environments such as clinical and physical activity laboratories for the assessment of mobility. Inertial measurement units (IMUs) are used as to monitor the motion of human movement for prolonged periods of time and without space limitations. The focus in this study was to provide a robust, low-cost and an unobtrusive solution for evaluating human motion using a single IMU. First part of the study focused on monitoring and classification of the daily life activities. A simple method that analyses the variations in signal was developed to distinguish two types of activity intervals: active and inactive. Neural classifier was used to classify active intervals; the angle with respect to gravity was used to classify inactive intervals. Second part of the study focused on extraction of gait parameters using a single inertial measurement unit (IMU) attached to the pelvis. Two complementary methods were proposed for gait parameters estimation. First method was a wavelet based method developed for the estimation of gait events. Second method was developed for estimating step and stride length during level walking using the estimations of the previous method. A special integration algorithm was extended to operate on each gait cycle using a specially designed Kalman filter. The developed methods were also applied on various scenarios. Activity monitoring method was used in a PRIN’07 project to assess the mobility levels of individuals living in a urban area. The same method was applied on volleyball players to analyze the fitness levels of them by monitoring their daily life activities. The methods proposed in these studies provided a simple, unobtrusive and low-cost solution for monitoring and assessing activities outside of controlled environments.

Lesion detection and classification in breast cancer: evaluation of approaches based on morphological features, tracer kinetic modelling and semi-quantitative parameters in MR functional imaging (DCE-MRI)

The diagnosis, grading and classification of tumours has benefited considerably from the development of DCE-MRI which is now essential to the adequate clinical management of many tumour types due to its capability in detecting active angiogenesis. Several strategies have been proposed for DCE-MRI evaluation. Visual inspection of contrast agent concentration curves vs time is a very simple yet operator dependent procedure, therefore more objective approaches have been developed in order to facilitate comparison between studies. In so called model free approaches, descriptive or heuristic information extracted from time series raw data have been used for tissue classification. The main issue concerning these schemes is that they have not a direct interpretation in terms of physiological properties of the tissues. On the other hand, model based investigations typically involve compartmental tracer kinetic modelling and pixel-by-pixel estimation of kinetic parameters via non-linear regression applied on region of interests opportunely selected by the physician. This approach has the advantage to provide parameters directly related to the pathophysiological properties of the tissue such as vessel permeability, local regional blood flow, extraction fraction, concentration gradient between plasma and extravascular-extracellular space. Anyway, nonlinear modelling is computational demanding and the accuracy of the estimates can be affected by the signal-to-noise ratio and by the initial solutions. The principal aim of this thesis is investigate the use of semi-quantitative and quantitative parameters for segmentation and classification of breast lesion. The objectives can be subdivided as follow: describe the principal techniques to evaluate time intensity curve in DCE-MRI with focus on kinetic model proposed in literature; to evaluate the influence in parametrization choice for a classic bi-compartmental kinetic models; to evaluate the performance of a method for simultaneous tracer kinetic modelling and pixel classification; to evaluate performance of machine learning techniques training for segmentation and classification of breast lesion.

Quantum Integrability in Non-Linear Sigma Models related to Gauge/String Correspondences

The Thermodynamic Bethe Ansatz analysis is carried out for the extended-CP^N class of integrable 2-dimensional Non-Linear Sigma Models related to the low energy limit of the AdS_4xCP^3 type IIA superstring theory. The principal aim of this program is to obtain further non-perturbative consistency check to the S-matrix proposed to describe the scattering processes between the fundamental excitations of the theory by analyzing the structure of the Renormalization Group flow. As a noteworthy byproduct we eventually obtain a novel class of TBA models which fits in the known classification but with several important differences. The TBA framework allows the evaluation of some exact quantities related to the conformal UV limit of the model: effective central charge, conformal dimension of the perturbing operator and field content of the underlying CFT. The knowledge of this physical quantities has led to the possibility of conjecturing a perturbed CFT realization of the integrable models in terms of coset Kac-Moody CFT. The set of numerical tools and programs developed ad hoc to solve the problem at hand is also discussed in some detail with references to the code.

Learning Methods and Algorithms for Semantic Text Classification across Multiple Domains

Information is nowadays a key resource: machine learning and data mining techniques have been developed to extract high-level information from great amounts of data. As most data comes in form of unstructured text in natural languages, research on text mining is currently very active and dealing with practical problems. Among these, text categorization deals with the automatic organization of large quantities of documents in priorly defined taxonomies of topic categories, possibly arranged in large hierarchies. In commonly proposed machine learning approaches, classifiers are automatically trained from pre-labeled documents: they can perform very accurate classification, but often require a consistent training set and notable computational effort. Methods for cross-domain text categorization have been proposed, allowing to leverage a set of labeled documents of one domain to classify those of another one. Most methods use advanced statistical techniques, usually involving tuning of parameters. A first contribution presented here is a method based on nearest centroid classification, where profiles of categories are generated from the known domain and then iteratively adapted to the unknown one. Despite being conceptually simple and having easily tuned parameters, this method achieves state-of-the-art accuracy in most benchmark datasets with fast running times. A second, deeper contribution involves the design of a domain-independent model to distinguish the degree and type of relatedness between arbitrary documents and topics, inferred from the different types of semantic relationships between respective representative words, identified by specific search algorithms. The application of this model is tested on both flat and hierarchical text categorization, where it potentially allows the efficient addition of new categories during classification. Results show that classification accuracy still requires improvements, but models generated from one domain are shown to be effectively able to be reused in a different one.

Tensor-Train decomposition for image classification problems

In these last years a great effort has been put in the development of new techniques for automatic object classification, also due to the consequences in many applications such as medical imaging or driverless cars. To this end, several mathematical models have been developed from logistic regression to neural networks. A crucial aspect of these so called classification algorithms is the use of algebraic tools to represent and approximate the input data. In this thesis, we examine two different models for image classification based on a particular tensor decomposition named Tensor-Train (TT) decomposition. The use of tensor approaches preserves the multidimensional structure of the data and the neighboring relations among pixels. Furthermore the Tensor-Train, differently from other tensor decompositions, does not suffer from the curse of dimensionality making it an extremely powerful strategy when dealing with high-dimensional data. It also allows data compression when combined with truncation strategies that reduce memory requirements without spoiling classification performance. The first model we propose is based on a direct decomposition of the database by means of the TT decomposition to find basis vectors used to classify a new object. The second model is a tensor dictionary learning model, based on the TT decomposition where the terms of the decomposition are estimated using a proximal alternating linearized minimization algorithm with a spectral stepsize.

Diffusive models and chaos indicators for non-linear betatron motion in circular hadron accelerators

Understanding the complex dynamics of beam-halo formation and evolution in circular particle accelerators is crucial for the design of current and future rings, particularly those utilizing superconducting magnets such as the CERN Large Hadron Collider (LHC), its luminosity upgrade HL-LHC, and the proposed Future Circular Hadron Collider (FCC-hh). A recent diffusive framework, which describes the evolution of the beam distribution by means of a Fokker-Planck equation, with diffusion coefficient derived from the Nekhoroshev theorem, has been proposed to describe the long-term behaviour of beam dynamics and particle losses. In this thesis, we discuss the theoretical foundations of this framework, and propose the implementation of an original measurement protocol based on collimator scans in view of measuring the Nekhoroshev-like diffusive coefficient by means of beam loss data. The available LHC collimator scan data, unfortunately collected without the proposed measurement protocol, have been successfully analysed using the proposed framework. This approach is also applied to datasets from detailed measurements of the impact on the beam losses of so-called long-range beam-beam compensators also at the LHC. Furthermore, dynamic indicators have been studied as a tool for exploring the phase-space properties of realistic accelerator lattices in single-particle tracking simulations. By first examining the classification performance of known and new indicators in detecting the chaotic character of initial conditions for a modulated Hénon map and then applying this knowledge to study the properties of realistic accelerator lattices, we tried to identify a connection between the presence of chaotic regions in the phase space and Nekhoroshev-like diffusive behaviour, providing new tools to the accelerator physics community.

Digital 3D reconstruction as a research environment in art and architecture history: uncertainty classification and visualisation

The dissertation addresses the still not solved challenges concerned with the source-based digital 3D reconstruction, visualisation and documentation in the domain of archaeology, art and architecture history. The emerging BIM methodology and the exchange data format IFC are changing the way of collaboration, visualisation and documentation in the planning, construction and facility management process. The introduction and development of the Semantic Web (Web 3.0), spreading the idea of structured, formalised and linked data, offers semantically enriched human- and machine-readable data. In contrast to civil engineering and cultural heritage, academic object-oriented disciplines, like archaeology, art and architecture history, are acting as outside spectators. Since the 1990s, it has been argued that a 3D model is not likely to be considered a scientific reconstruction unless it is grounded on accurate documentation and visualisation. However, these standards are still missing and the validation of the outcomes is not fulfilled. Meanwhile, the digital research data remain ephemeral and continue to fill the growing digital cemeteries. This study focuses, therefore, on the evaluation of the source-based digital 3D reconstructions and, especially, on uncertainty assessment in the case of hypothetical reconstructions of destroyed or never built artefacts according to scientific principles, making the models shareable and reusable by a potentially wide audience. The work initially focuses on terminology and on the definition of a workflow especially related to the classification and visualisation of uncertainty. The workflow is then applied to specific cases of 3D models uploaded to the DFG repository of the AI Mainz. In this way, the available methods of documenting, visualising and communicating uncertainty are analysed. In the end, this process will lead to a validation or a correction of the workflow and the initial assumptions, but also (dealing with different hypotheses) to a better definition of the levels of uncertainty.

Efficient and low-computational predictive models for spectral sensors

Spectral sensors are a wide class of devices that are extremely useful for detecting essential information of the environment and materials with high degree of selectivity. Recently, they have achieved high degrees of integration and low implementation cost to be suited for fast, small, and non-invasive monitoring systems. However, the useful information is hidden in spectra and it is difficult to decode. So, mathematical algorithms are needed to infer the value of the variables of interest from the acquired data. Between the different families of predictive modeling, Principal Component Analysis and the techniques stemmed from it can provide very good performances, as well as small computational and memory requirements. For these reasons, they allow the implementation of the prediction even in embedded and autonomous devices. In this thesis, I will present 4 practical applications of these algorithms to the prediction of different variables: moisture of soil, moisture of concrete, freshness of anchovies/sardines, and concentration of gasses. In all of these cases, the workflow will be the same. Initially, an acquisition campaign was performed to acquire both spectra and the variables of interest from samples. Then these data are used as input for the creation of the prediction models, to solve both classification and regression problems. From these models, an array of calibration coefficients is derived and used for the implementation of the prediction in an embedded system. The presented results will show that this workflow was successfully applied to very different scientific fields, obtaining autonomous and non-invasive devices able to predict the value of physical parameters of choice from new spectral acquisitions.

Mind the gaps: cognitive-inspired AI for high-level visual sensemaking. Towards abstract concept image classification

The abundance of visual data and the push for robust AI are driving the need for automated visual sensemaking. Computer Vision (CV) faces growing demand for models that can discern not only what images "represent," but also what they "evoke." This is a demand for tools mimicking human perception at a high semantic level, categorizing images based on concepts like freedom, danger, or safety. However, automating this process is challenging due to entropy, scarcity, subjectivity, and ethical considerations. These challenges not only impact performance but also underscore the critical need for interoperability. This dissertation focuses on abstract concept-based (AC) image classification, guided by three technical principles: situated grounding, performance enhancement, and interpretability. We introduce ART-stract, a novel dataset of cultural images annotated with ACs, serving as the foundation for a series of experiments across four key domains: assessing the effectiveness of the end-to-end DL paradigm, exploring cognitive-inspired semantic intermediaries, incorporating cultural and commonsense aspects, and neuro-symbolic integration of sensory-perceptual data with cognitive-based knowledge. Our results demonstrate that integrating CV approaches with semantic technologies yields methods that surpass the current state of the art in AC image classification, outperforming the end-to-end deep vision paradigm. The results emphasize the role semantic technologies can play in developing both effective and interpretable systems, through the capturing, situating, and reasoning over knowledge related to visual data. Furthermore, this dissertation explores the complex interplay between technical and socio-technical factors. By merging technical expertise with an understanding of human and societal aspects, we advocate for responsible labeling and training practices in visual media. These insights and techniques not only advance efforts in CV and explainable artificial intelligence but also propel us toward an era of AI development that harmonizes technical prowess with deep awareness of its human and societal implications.