Applying machine learning: a multi-role perspective

Machine (and deep) learning technologies are more and more present in several fields. It is undeniable that many aspects of our society are empowered by such technologies: web searches, content filtering on social networks, recommendations on e-commerce websites, mobile applications, etc., in addition to academic research. Moreover, mobile devices and internet sites, e.g., social networks, support the collection and sharing of information in real time. The pervasive deployment of the aforementioned technological instruments, both hardware and software, has led to the production of huge amounts of data. Such data has become more and more unmanageable, posing challenges to conventional computing platforms, and paving the way to the development and widespread use of the machine and deep learning. Nevertheless, machine learning is not only a technology. Given a task, machine learning is a way of proceeding (a way of thinking), and as such can be approached from different perspectives (points of view). This, in particular, will be the focus of this research. The entire work concentrates on machine learning, starting from different sources of data, e.g., signals and images, applied to different domains, e.g., Sport Science and Social History, and analyzed from different perspectives: from a non-data scientist point of view through tools and platforms; setting a problem stage from scratch; implementing an effective application for classification tasks; improving user interface experience through Data Visualization and eXtended Reality. In essence, not only in a quantitative task, not only in a scientific environment, and not only from a data-scientist perspective, machine (and deep) learning can do the difference.

Data surveillance for infectious diseases: models, complex networks and machine learning

In this thesis, we investigate the role of applied physics in epidemiological surveillance through the application of mathematical models, network science and machine learning. The spread of a communicable disease depends on many biological, social, and health factors. The large masses of data available make it possible, on the one hand, to monitor the evolution and spread of pathogenic organisms; on the other hand, to study the behavior of people, their opinions and habits. Presented here are three lines of research in which an attempt was made to solve real epidemiological problems through data analysis and the use of statistical and mathematical models. In Chapter 1, we applied language-inspired Deep Learning models to transform influenza protein sequences into vectors encoding their information content. We then attempted to reconstruct the antigenic properties of different viral strains using regression models and to identify the mutations responsible for vaccine escape. In Chapter 2, we constructed a compartmental model to describe the spread of a bacterium within a hospital ward. The model was informed and validated on time series of clinical measurements, and a sensitivity analysis was used to assess the impact of different control measures. Finally (Chapter 3) we reconstructed the network of retweets among COVID-19 themed Twitter users in the early months of the SARS-CoV-2 pandemic. By means of community detection algorithms and centrality measures, we characterized users’ attention shifts in the network, showing that scientific communities, initially the most retweeted, lost influence over time to national political communities. In the Conclusion, we highlighted the importance of the work done in light of the main contemporary challenges for epidemiological surveillance. In particular, we present reflections on the importance of nowcasting and forecasting, the relationship between data and scientific research, and the need to unite the different scales of epidemiological surveillance.

Developing a radiomic based machine learning model to identify patients with resected non-small-cell lung cancer at high risk of relapse

Background There is a wide variation of recurrence risk of Non-small-cell lung cancer (NSCLC) within the same Tumor Node Metastasis (TNM) stage, suggesting that other parameters are involved in determining this probability. Radiomics allows extraction of quantitative information from images that can be used for clinical purposes. The primary objective of this study is to develop a radiomic prognostic model that predicts a 3 year disease free-survival (DFS) of resected Early Stage (ES) NSCLC patients. Material and Methods 56 pre-surgery non contrast Computed Tomography (CT) scans were retrieved from the PACS of our institution and anonymized. Then they were automatically segmented with an open access deep learning pipeline and reviewed by an experienced radiologist to obtain 3D masks of the NSCLC. Images and masks underwent to resampling normalization and discretization. From the masks hundreds Radiomic Features (RF) were extracted using Py-Radiomics. Hence, RF were reduced to select the most representative features. The remaining RF were used in combination with Clinical parameters to build a DFS prediction model using Leave-one-out cross-validation (LOOCV) with Random Forest. Results and Conclusion A poor agreement between the radiologist and the automatic segmentation algorithm (DICE score of 0.37) was found. Therefore, another experienced radiologist manually segmented the lesions and only stable and reproducible RF were kept. 50 RF demonstrated a high correlation with the DFS but only one was confirmed when clinicopathological covariates were added: Busyness a Neighbouring Gray Tone Difference Matrix (HR 9.610). 16 clinical variables (which comprised TNM) were used to build the LOOCV model demonstrating a higher Area Under the Curve (AUC) when RF were included in the analysis (0.67 vs 0.60) but the difference was not statistically significant (p=0,5147).

Machine learning methods for prediction of disulphide bonding states of cysteine residues in proteins

The goal of this thesis work is to develop a computational method based on machine learning techniques for predicting disulfide-bonding states of cysteine residues in proteins, which is a sub-problem of a bigger and yet unsolved problem of protein structure prediction. Improvement in the prediction of disulfide bonding states of cysteine residues will help in putting a constraint in the three dimensional (3D) space of the respective protein structure, and thus will eventually help in the prediction of 3D structure of proteins. Results of this work will have direct implications in site-directed mutational studies of proteins, proteins engineering and the problem of protein folding. We have used a combination of Artificial Neural Network (ANN) and Hidden Markov Model (HMM), the so-called Hidden Neural Network (HNN) as a machine learning technique to develop our prediction method. By using different global and local features of proteins (specifically profiles, parity of cysteine residues, average cysteine conservation, correlated mutation, sub-cellular localization, and signal peptide) as inputs and considering Eukaryotes and Prokaryotes separately we have reached to a remarkable accuracy of 94% on cysteine basis for both Eukaryotic and Prokaryotic datasets, and an accuracy of 90% and 93% on protein basis for Eukaryotic dataset and Prokaryotic dataset respectively. These accuracies are best so far ever reached by any existing prediction methods, and thus our prediction method has outperformed all the previously developed approaches and therefore is more reliable. Most interesting part of this thesis work is the differences in the prediction performances of Eukaryotes and Prokaryotes at the basic level of input coding when ‘profile’ information was given as input to our prediction method. And one of the reasons for this we discover is the difference in the amino acid composition of the local environment of bonded and free cysteine residues in Eukaryotes and Prokaryotes. Eukaryotic bonded cysteine examples have a ‘symmetric-cysteine-rich’ environment, where as Prokaryotic bonded examples lack it.

Machine-learning methods for structure prediction of β-barrel membrane proteins

Different types of proteins exist with diverse functions that are essential for living organisms. An important class of proteins is represented by transmembrane proteins which are specifically designed to be inserted into biological membranes and devised to perform very important functions in the cell such as cell communication and active transport across the membrane. Transmembrane β-barrels (TMBBs) are a sub-class of membrane proteins largely under-represented in structure databases because of the extreme difficulty in experimental structure determination. For this reason, computational tools that are able to predict the structure of TMBBs are needed. In this thesis, two computational problems related to TMBBs were addressed: the detection of TMBBs in large datasets of proteins and the prediction of the topology of TMBB proteins. Firstly, a method for TMBB detection was presented based on a novel neural network framework for variable-length sequence classification. The proposed approach was validated on a non-redundant dataset of proteins. Furthermore, we carried-out genome-wide detection using the entire Escherichia coli proteome. In both experiments, the method significantly outperformed other existing state-of-the-art approaches, reaching very high PPV (92%) and MCC (0.82). Secondly, a method was also introduced for TMBB topology prediction. The proposed approach is based on grammatical modelling and probabilistic discriminative models for sequence data labeling. The method was evaluated using a newly generated dataset of 38 TMBB proteins obtained from high-resolution data in the PDB. Results have shown that the model is able to correctly predict topologies of 25 out of 38 protein chains in the dataset. When tested on previously released datasets, the performances of the proposed approach were measured as comparable or superior to the current state-of-the-art of TMBB topology prediction.

Towards Unstructured Knowledge Integration in Natural Language Processing

In the last decades, Artificial Intelligence has witnessed multiple breakthroughs in deep learning. In particular, purely data-driven approaches have opened to a wide variety of successful applications due to the large availability of data. Nonetheless, the integration of prior knowledge is still required to compensate for specific issues like lack of generalization from limited data, fairness, robustness, and biases. In this thesis, we analyze the methodology of integrating knowledge into deep learning models in the field of Natural Language Processing (NLP). We start by remarking on the importance of knowledge integration. We highlight the possible shortcomings of these approaches and investigate the implications of integrating unstructured textual knowledge. We introduce Unstructured Knowledge Integration (UKI) as the process of integrating unstructured knowledge into machine learning models. We discuss UKI in the field of NLP, where knowledge is represented in a natural language format. We identify UKI as a complex process comprised of multiple sub-processes, different knowledge types, and knowledge integration properties to guarantee. We remark on the challenges of integrating unstructured textual knowledge and bridge connections with well-known research areas in NLP. We provide a unified vision of structured knowledge extraction (KE) and UKI by identifying KE as a sub-process of UKI. We investigate some challenging scenarios where structured knowledge is not a feasible prior assumption and formulate each task from the point of view of UKI. We adopt simple yet effective neural architectures and discuss the challenges of such an approach. Finally, we identify KE as a form of symbolic representation. From this perspective, we remark on the need of defining sophisticated UKI processes to verify the validity of knowledge integration. To this end, we foresee frameworks capable of combining symbolic and sub-symbolic representations for learning as a solution.

The three-dimensional single-bin-size bin packing problem: combining metaheuristic and machine learning approaches

The Three-Dimensional Single-Bin-Size Bin Packing Problem is one of the most studied problem in the Cutting & Packing category. From a strictly mathematical point of view, it consists of packing a finite set of strongly heterogeneous “small” boxes, called items, into a finite set of identical “large” rectangles, called bins, minimizing the unused volume and requiring that the items are packed without overlapping. The great interest is mainly due to the number of real-world applications in which it arises, such as pallet and container loading, cutting objects out of a piece of material and packaging design. Depending on these real-world applications, more objective functions and more practical constraints could be needed. After a brief discussion about the real-world applications of the problem and a exhaustive literature review, the design of a two-stage algorithm to solve the aforementioned problem is presented. The algorithm must be able to provide the spatial coordinates of the placed boxes vertices and also the optimal boxes input sequence, while guaranteeing geometric, stability, fragility constraints and a reduced computational time. Due to NP-hard complexity of this type of combinatorial problems, a fusion of metaheuristic and machine learning techniques is adopted. In particular, a hybrid genetic algorithm coupled with a feedforward neural network is used. In the first stage, a rich dataset is created starting from a set of real input instances provided by an industrial company and the feedforward neural network is trained on it. After its training, given a new input instance, the hybrid genetic algorithm is able to run using the neural network output as input parameter vector, providing as output the optimal solution. The effectiveness of the proposed works is confirmed via several experimental tests.

Development of unsupervised learning methods with applications to life sciences data

Machine Learning makes computers capable of performing tasks typically requiring human intelligence. A domain where it is having a considerable impact is the life sciences, allowing to devise new biological analysis protocols, develop patients’ treatments efficiently and faster, and reduce healthcare costs. This Thesis work presents new Machine Learning methods and pipelines for the life sciences focusing on the unsupervised field. At a methodological level, two methods are presented. The first is an “Ab Initio Local Principal Path” and it is a revised and improved version of a pre-existing algorithm in the manifold learning realm. The second contribution is an improvement over the Import Vector Domain Description (one-class learning) through the Kullback-Leibler divergence. It hybridizes kernel methods to Deep Learning obtaining a scalable solution, an improved probabilistic model, and state-of-the-art performances. Both methods are tested through several experiments, with a central focus on their relevance in life sciences. Results show that they improve the performances achieved by their previous versions. At the applicative level, two pipelines are presented. The first one is for the analysis of RNA-Seq datasets, both transcriptomic and single-cell data, and is aimed at identifying genes that may be involved in biological processes (e.g., the transition of tissues from normal to cancer). In this project, an R package is released on CRAN to make the pipeline accessible to the bioinformatic Community through high-level APIs. The second pipeline is in the drug discovery domain and is useful for identifying druggable pockets, namely regions of a protein with a high probability of accepting a small molecule (a drug). Both these pipelines achieve remarkable results. Lastly, a detour application is developed to identify the strengths/limitations of the “Principal Path” algorithm by analyzing Convolutional Neural Networks induced vector spaces. This application is conducted in the music and visual arts domains.

Variational and learning models for image and time series inverse problems

Inverse problems are at the core of many challenging applications. Variational and learning models provide estimated solutions of inverse problems as the outcome of specific reconstruction maps. In the variational approach, the result of the reconstruction map is the solution of a regularized minimization problem encoding information on the acquisition process and prior knowledge on the solution. In the learning approach, the reconstruction map is a parametric function whose parameters are identified by solving a minimization problem depending on a large set of data. In this thesis, we go beyond this apparent dichotomy between variational and learning models and we show they can be harmoniously merged in unified hybrid frameworks preserving their main advantages. We develop several highly efficient methods based on both these model-driven and data-driven strategies, for which we provide a detailed convergence analysis. The arising algorithms are applied to solve inverse problems involving images and time series. For each task, we show the proposed schemes improve the performances of many other existing methods in terms of both computational burden and quality of the solution. In the first part, we focus on gradient-based regularized variational models which are shown to be effective for segmentation purposes and thermal and medical image enhancement. We consider gradient sparsity-promoting regularized models for which we develop different strategies to estimate the regularization strength. Furthermore, we introduce a novel gradient-based Plug-and-Play convergent scheme considering a deep learning based denoiser trained on the gradient domain. In the second part, we address the tasks of natural image deblurring, image and video super resolution microscopy and positioning time series prediction, through deep learning based methods. We boost the performances of supervised, such as trained convolutional and recurrent networks, and unsupervised deep learning strategies, such as Deep Image Prior, by penalizing the losses with handcrafted regularization terms.

Computational methods applied to undeciphered scripts from the Aegean and Cyprus: Linear A and Cypro-Minoan

The study of ancient, undeciphered scripts presents unique challenges, that depend both on the nature of the problem and on the peculiarities of each writing system. In this thesis, I present two computational approaches that are tailored to two different tasks and writing systems. The first of these methods is aimed at the decipherment of the Linear A afraction signs, in order to discover their numerical values. This is achieved with a combination of constraint programming, ad-hoc metrics and paleographic considerations. The second main contribution of this thesis regards the creation of an unsupervised deep learning model which uses drawings of signs from ancient writing system to learn to distinguish different graphemes in the vector space. This system, which is based on techniques used in the field of computer vision, is adapted to the study of ancient writing systems by incorporating information about sequences in the model, mirroring what is often done in natural language processing. In order to develop this model, the Cypriot Greek Syllabary is used as a target, since this is a deciphered writing system. Finally, this unsupervised model is adapted to the undeciphered Cypro-Minoan and it is used to answer open questions about this script. In particular, by reconstructing multiple allographs that are not agreed upon by paleographers, it supports the idea that Cypro-Minoan is a single script and not a collection of three script like it was proposed in the literature. These results on two different tasks shows that computational methods can be applied to undeciphered scripts, despite the relatively low amount of available data, paving the way for further advancement in paleography using these methods.

Automotive Radars:towards smarter vehicles

Radars are expected to become the main sensors in various civilian applications, especially for autonomous driving. Their success is mainly due to the availability of low cost integrated devices, equipped with compact antenna arrays, and computationally efficient signal processing techniques. This thesis focuses on the study and the development of different deterministic and learning based techniques for colocated multiple-input multiple-output (MIMO) radars. In particular, after providing an overview on the architecture of these devices, the problem of detecting and estimating multiple targets in stepped frequency continuous wave (SFCW) MIMO radar systems is investigated and different deterministic techniques solving it are illustrated. Moreover, novel solutions, based on an approximate maximum likelihood approach, are developed. The accuracy achieved by all the considered algorithms is assessed on the basis of the raw data acquired from low power wideband radar devices. The results demonstrate that the developed algorithms achieve reasonable accuracies, but at the price of different computational efforts. Another important technical problem investigated in this thesis concerns the exploitation of machine learning and deep learning techniques in the field of colocated MIMO radars. In this thesis, after providing a comprehensive overview of the machine learning and deep learning techniques currently being considered for use in MIMO radar systems, their performance in two different applications is assessed on the basis of synthetically generated and experimental datasets acquired through a commercial frequency modulated continuous wave (FMCW) MIMO radar. Finally, the application of colocated MIMO radars to autonomous driving in smart agriculture is illustrated.

Monitoring and Prediction of Thermal Emergencies in High Performance Computing Systems

Modern scientific discoveries are driven by an unsatisfiable demand for computational resources. High-Performance Computing (HPC) systems are an aggregation of computing power to deliver considerably higher performance than one typical desktop computer can provide, to solve large problems in science, engineering, or business. An HPC room in the datacenter is a complex controlled environment that hosts thousands of computing nodes that consume electrical power in the range of megawatts, which gets completely transformed into heat. Although a datacenter contains sophisticated cooling systems, our studies indicate quantitative evidence of thermal bottlenecks in real-life production workload, showing the presence of significant spatial and temporal thermal and power heterogeneity. Therefore minor thermal issues/anomalies can potentially start a chain of events that leads to an unbalance between the amount of heat generated by the computing nodes and the heat removed by the cooling system originating thermal hazards. Although thermal anomalies are rare events, anomaly detection/prediction in time is vital to avoid IT and facility equipment damage and outage of the datacenter, with severe societal and business losses. For this reason, automated approaches to detect thermal anomalies in datacenters have considerable potential. This thesis analyzed and characterized the power and thermal characteristics of a Tier0 datacenter (CINECA) during production and under abnormal thermal conditions. Then, a Deep Learning (DL)-powered thermal hazard prediction framework is proposed. The proposed models are validated against real thermal hazard events reported for the studied HPC cluster while in production. This thesis is the first empirical study of thermal anomaly detection and prediction techniques of a real large-scale HPC system to the best of my knowledge. For this thesis, I used a large-scale dataset, monitoring data of tens of thousands of sensors for around 24 months with a data collection rate of around 20 seconds.

Interpretable Convolutional Neural Networks for Decoding and Analyzing Neural Time Series Data

Machine learning is widely adopted to decode multi-variate neural time series, including electroencephalographic (EEG) and single-cell recordings. Recent solutions based on deep learning (DL) outperformed traditional decoders by automatically extracting relevant discriminative features from raw or minimally pre-processed signals. Convolutional Neural Networks (CNNs) have been successfully applied to EEG and are the most common DL-based EEG decoders in the state-of-the-art (SOA). However, the current research is affected by some limitations. SOA CNNs for EEG decoding usually exploit deep and heavy structures with the risk of overfitting small datasets, and architectures are often defined empirically. Furthermore, CNNs are mainly validated by designing within-subject decoders. Crucially, the automatically learned features mainly remain unexplored; conversely, interpreting these features may be of great value to use decoders also as analysis tools, highlighting neural signatures underlying the different decoded brain or behavioral states in a data-driven way. Lastly, SOA DL-based algorithms used to decode single-cell recordings rely on more complex, slower to train and less interpretable networks than CNNs, and the use of CNNs with these signals has not been investigated. This PhD research addresses the previous limitations, with reference to P300 and motor decoding from EEG, and motor decoding from single-neuron activity. CNNs were designed light, compact, and interpretable. Moreover, multiple training strategies were adopted, including transfer learning, which could reduce training times promoting the application of CNNs in practice. Furthermore, CNN-based EEG analyses were proposed to study neural features in the spatial, temporal and frequency domains, and proved to better highlight and enhance relevant neural features related to P300 and motor states than canonical EEG analyses. Remarkably, these analyses could be used, in perspective, to design novel EEG biomarkers for neurological or neurodevelopmental disorders. Lastly, CNNs were developed to decode single-neuron activity, providing a better compromise between performance and model complexity.

Big Data approaches as a support for precision livestock farming techniques

With the advent of new technologies it is increasingly easier to find data of different nature from even more accurate sensors that measure the most disparate physical quantities and with different methodologies. The collection of data thus becomes progressively important and takes the form of archiving, cataloging and online and offline consultation of information. Over time, the amount of data collected can become so relevant that it contains information that cannot be easily explored manually or with basic statistical techniques. The use of Big Data therefore becomes the object of more advanced investigation techniques, such as Machine Learning and Deep Learning. In this work some applications in the world of precision zootechnics and heat stress accused by dairy cows are described. Experimental Italian and German stables were involved for the training and testing of the Random Forest algorithm, obtaining a prediction of milk production depending on the microclimatic conditions of the previous days with satisfactory accuracy. Furthermore, in order to identify an objective method for identifying production drops, compared to the Wood model, typically used as an analytical model of the lactation curve, a Robust Statistics technique was used. Its application on some sample lactations and the results obtained allow us to be confident about the use of this method in the future.

Deploying and processing neural representations of signals

Neural representations (NR) have emerged in the last few years as a powerful tool to represent signals from several domains, such as images, 3D shapes, or audio. Indeed, deep neural networks have been shown capable of approximating continuous functions that describe a given signal with theoretical infinite resolution. This finding allows obtaining representations whose memory footprint is fixed and decoupled from the resolution at which the underlying signal can be sampled, something that is not possible with traditional discrete representations, e.g., grids of pixels for images or voxels for 3D shapes. During the last two years, many techniques have been proposed to improve the capability of NR to approximate high-frequency details and to make the optimization procedures required to obtain NR less demanding both in terms of time and data requirements, motivating many researchers to deploy NR as the main form of data representation for complex pipelines. Following this line of research, we first show that NR can approximate precisely Unsigned Distance Functions, providing an effective way to represent garments that feature open 3D surfaces and unknown topology. Then, we present a pipeline to obtain in a few minutes a compact Neural Twin® for a given object, by exploiting the recent advances in modeling neural radiance fields. Furthermore, we move a step in the direction of adopting NR as a standalone representation, by considering the possibility of performing downstream tasks by processing directly the NR weights. We first show that deep neural networks can be compressed into compact latent codes. Then, we show how this technique can be exploited to perform deep learning on implicit neural representations (INR) of 3D shapes, by only looking at the weights of the networks.