A Distributed Support based on Kubeflow to Ease MLOps of Federated Learning Algorithms

The main objective of my thesis work is to exploit the Google native and open-source platform Kubeflow, specifically using Kubeflow pipelines, to execute a Federated Learning scalable ML process in a 5G-like and simplified test architecture hosting a Kubernetes cluster and apply the largely adopted FedAVG algorithm and FedProx its optimization empowered by the ML platform ‘s abilities to ease the development and production cycle of this specific FL process. FL algorithms are more are and more promising and adopted both in Cloud application development and 5G communication enhancement through data coming from the monitoring of the underlying telco infrastructure and execution of training and data aggregation at edge nodes to optimize the global model of the algorithm ( that could be used for example for resource provisioning to reach an agreed QoS for the underlying network slice) and after a study and a research over the available papers and scientific articles related to FL with the help of the CTTC that suggests me to study and use Kubeflow to bear the algorithm we found out that this approach for the whole FL cycle deployment was not documented and may be interesting to investigate more in depth. This study may lead to prove the efficiency of the Kubeflow platform itself for this need of development of new FL algorithms that will support new Applications and especially test the FedAVG algorithm performances in a simulated client to cloud communication using a MNIST dataset for FL as benchmark.

Machine learning "as a service" for high energy physics (MLaaS4HEP): evolution of a framework for ML-based physics analyses

The scientific success of the LHC experiments at CERN highly depends on the availability of computing resources which efficiently store, process, and analyse the amount of data collected every year. This is ensured by the Worldwide LHC Computing Grid infrastructure that connect computing centres distributed all over the world with high performance network. LHC has an ambitious experimental program for the coming years, which includes large investments and improvements both for the hardware of the detectors and for the software and computing systems, in order to deal with the huge increase in the event rate expected from the High Luminosity LHC (HL-LHC) phase and consequently with the huge amount of data that will be produced. Since few years the role of Artificial Intelligence has become relevant in the High Energy Physics (HEP) world. Machine Learning (ML) and Deep Learning algorithms have been successfully used in many areas of HEP, like online and offline reconstruction programs, detector simulation, object reconstruction, identification, Monte Carlo generation, and surely they will be crucial in the HL-LHC phase. This thesis aims at contributing to a CMS R&D project, regarding a ML "as a Service" solution for HEP needs (MLaaS4HEP). It consists in a data-service able to perform an entire ML pipeline (in terms of reading data, processing data, training ML models, serving predictions) in a completely model-agnostic fashion, directly using ROOT files of arbitrary size from local or distributed data sources. This framework has been updated adding new features in the data preprocessing phase, allowing more flexibility to the user. Since the MLaaS4HEP framework is experiment agnostic, the ATLAS Higgs Boson ML challenge has been chosen as physics use case, with the aim to test MLaaS4HEP and the contribution done with this work.

Neural Network Based UAV Fault Tolerant Control with Aerodynamic Disturbance Estimation and Compensation

In this thesis, the problem of controlling a quadrotor UAV is considered. It is done by presenting an original control system, designed as a combination of Neural Networks and Disturbance Observer, using a composite learning approach for a system of the second order, which is a novel methodology in literature. After a brief introduction about the quadrotors, the concepts needed to understand the controller are presented, such as the main notions of advanced control, the basic structure and design of a Neural Network, the modeling of a quadrotor and its dynamics. The full simulator, developed on the MATLAB Simulink environment, used throughout the whole thesis, is also shown. For the guidance and control purposes, a Sliding Mode Controller, used as a reference, it is firstly introduced, and its theory and implementation on the simulator are illustrated. Finally the original controller is introduced, through its novel formulation, and implementation on the model. The effectiveness and robustness of the two controllers are then proven by extensive simulations in all different conditions of external disturbance and faults.

Geometric Deep Learning per il denoising di mesh 3D

La crescente disponibilità di scanner 3D ha reso più semplice l’acquisizione di modelli 3D dall’ambiente. A causa delle inevitabili imperfezioni ed errori che possono avvenire durante la fase di scansione, i modelli acquisiti possono risultare a volte inutilizzabili ed affetti da rumore. Le tecniche di denoising hanno come obiettivo quello di rimuovere dalla superficie della mesh 3D scannerizzata i disturbi provocati dal rumore, ristabilendo le caratteristiche originali della superficie senza introdurre false informazioni. Per risolvere questo problema, un approccio innovativo è quello di utilizzare il Geometric Deep Learning per addestrare una Rete Neurale in maniera da renderla in grado di eseguire efficacemente il denoising di mesh. L’obiettivo di questa tesi è descrivere il Geometric Deep Learning nell’ambito del problema sotto esame.

Machine Learning in Applicazione alla Prevenzione del Morbo di Alzheimer

Il morbo di Alzheimer è ancora una malattia incurabile. Negli ultimi anni l'aumento progressivo dell'aspettativa di vita ha contribuito a un'insorgenza maggiore di questa patologia, specialmente negli stati con l'età media più alta, tra cui l'Italia. La prevenzione risulta una delle poche vie con cui è possibile arginarne lo sviluppo, ed in questo testo vengono analizzate le potenzialità di alcune tecniche di Machine Learning atte alla creazione di modelli di supporto diagnostico per Alzheimer. Dopo un'opportuna introduzione al morbo di Alzheimer ed al funzionamento generale del Machine Learning, vengono presentate e approfondite due delle tecniche più promettenti per la diagnosi di patologie neurologiche, ovvero la Support Vector Machine (macchina a supporto vettoriale, SVM) e la Convolutional Neural Network (rete neurale convoluzionale, CNN), con annessi risultati, punti di forza e principali debolezze. La conclusione verterà sul possibile futuro delle intelligenze artificiali, con particolare attenzione all'ambito sanitario, e verranno discusse le principali difficoltà nelle quali queste incombono prima di essere commercializzate, insieme a plausibili soluzioni.

Trajectory Planning in UAV-Aided Wireless Networks Via Reinforcement Learning

This thesis is focused on the design of a flexible, dynamic and innovative telecommunication's system for future 6G applications on vehicular communications. The system is based on the development of drones acting as mobile base stations in an urban scenario to cope with the increasing traffic demand and avoid network's congestion conditions. In particular, the exploitation of Reinforcement Learning algorithms is used to let the drone learn autonomously how to behave in a scenario full of obstacles with the goal of tracking and serve the maximum number of moving vehicles, by at the same time, minimizing the energy consumed to perform its tasks. This project is an extraordinary opportunity to open the doors to a new way of applying and develop telecommunications in an urban scenario by mixing it to the rising world of the Artificial Intelligence.

Identificazione di caratteristiche rilevanti in capi di abbigliamento mediante deep-learning

Il mondo della moda è in continua e costante evoluzione, non solo dal punto di vista sociale, ma anche da quello tecnologico. Nel corso del presente elaborato si è studiata la possibilità di riconoscere e segmentare abiti presenti in una immagine utilizzando reti neurali profonde e approcci moderni. Sono state, quindi, analizzate reti quali FasterRCNN, MaskRCNN, YOLOv5, FashionPedia e Match-RCNN. In seguito si è approfondito l’addestramento delle reti neurali profonde in scenari di alta parallelizzazione e su macchine dotate di molteplici GPU al fine di ridurre i tempi di addestramento. Inoltre si è sperimentata la possibilità di creare una rete per prevedere se un determinato abito possa avere successo in futuro analizzando semplicemente dati passati e una immagine del vestito in questione. Necessaria per tali compiti è stata, inoltre, una approfondita analisi dei dataset esistenti nel mondo della moda e dei metodi per utilizzarli per l’addestramento. Il presente elaborato è stato svolto nell’ambito del progetto FA.RE.TRA. per il quale l'Università di Bologna svolge un compito di consulenza per lo studio di fattibilità su reti neurali in grado di svolgere i compiti menzionati.

Convolutional Neural Network Architectures for Template Matching

The Neural Networks customized and tested in this thesis (WaldoNet, FlowNet and PatchNet) are a first exploration and approach to the Template Matching task. The possibilities of extension are therefore many and some are proposed below. During my thesis, I have analyzed the functioning of the classical algorithms and adapted with deep learning algorithms. The features extracted from both the template and the query images resemble the keypoints of the SIFT algorithm. Then, instead of similarity function or keypoints matching, WaldoNet and PatchNet use the convolutional layer to compare the features, while FlowNet uses the correlational layer. In addition, I have identified the major challenges of the Template Matching task (affine/non-affine transformations, intensity changes...) and solved them with a careful design of the dataset.

A deep learning model for controlling the galvanizing process in a production line

In the industry of steelmaking, the process of galvanizing is a treatment which is applied to protect the steel from corrosion. The air knife effect (AKE) occurs when nozzles emit a steam of air on the surfaces of a steel strip to remove excess zinc from it. In our work we formalized the problem to control the AKE and we implemented, with the R&D dept.of MarcegagliaSPA, a DL model able to drive the AKE. We call it controller. It takes as input the tuple : a tuple of the physical conditions of the process line (t,h,s) with the target value of the zinc coating (c); and generates the expected tuple of (pres and dist) to drive the mechanical nozzles towards the (c). According to the requirements we designed the structure of the network. We collected and explored the data set of the historical data of the smart factory. Finally, we designed the loss function as sum of three components: the minimization between the coating addressed by the network and the target value we want to reach; and two weighted minimization components for both pressure and distance. In our solution we construct a second module, named coating net, to predict the coating of zinc resulting from the AKE when the conditions are applied to the prod. line. Its structure is made by a linear and a deep nonlinear “residual” component learned by empirical observations. The predictions made by the coating nets are used as ground truth in the loss function of the controller. By tuning the weights of the different components of the loss function, it is possible to train models with slightly different optimization purposes. In the tests we compared the regularization of different strategies with the standard one in condition of optimal estimation for both; the overall accuracy is ± 3 g/m^2 dal target for all of them. Lastly, we analyze how the controller modeled the current solutions with the new logic: the sub-optimal values of pres and dist can be optimize of 50% and 20%.

Comparison between 5G private network and Wifi6 in industrial environment

Industry 4.0 refers to the 4th industrial revolution and at its bases, we can see the digitalization and the automation of the assembly line. The whole production process has improved and evolved thanks to the advances made in networking, and AI studies, which include of course machine learning, cloud computing, IoT, and other technologies that are finally being implemented into the industrial scenario. All these technologies have in common a need for faster, more secure, robust, and reliable communication. One of the many solutions for these demands is the use of mobile communication technologies in the industrial environment, but which technology is better suited for these demands? Of course, the answer isn’t as simple as it seems. The 4th industrial revolution has a never seen incomparable potential with respect to the previous ones, every factory, enterprise, or company have different network demands, and even in each of these infrastructures, the demands may diversify by sector, or by application. For example, in the health care industry, there may be e a need for increased bandwidth for the analysis of high-definition videos or, faster speeds in order to have analytics occur in real-time, and again another application might be higher security and reliability to protect patients’ data. As seen above, choosing the right technology for the right environment and application, considers many things, and the ones just stated are but a speck of dust with respect to the overall picture. In this thesis, we will investigate a comparison between the use of two of the available technologies in use for the industrial environment: Wi-Fi 6 and 5G Private Networks in the specific case of a steel factory.

Deep Learning applicato alla diagnostica dei trasformatori

Il trasformatore è uno degli elementi più importanti di una rete di trasmissione; essendo il tramite fra reti di alta e media tensione, il suo corretto funzionamento garantisce l’alimentazione di tutti i dispositivi e carichi connessi alla linea. Oltre a questo, il trasformatore è anche l’elemento più costoso di tutta la linea elettrica; la sua manutenzione è di vitale importanza per evitare costi elevati per la sostituzione e disagi lungo la linea. Qui entra in gioco il ruolo della diagnostica; attraverso misure periodiche e mirate sul trasformatore è possibile agire tempestivamente ed evitare tutti i fenomeni precedentemente elencati. Nell’elaborato si tratterà l’analisi del trasformatore elettrico trifase durante il suo funzionamento, evidenziando i sottocomponenti e le rispettive criticità; inoltre, verranno mostrate le varie tecniche di diagnostica del trasformatore, in modo tale da poter estrarre un indice legato allo stato di vita, ossia l’Health Index. Ad oggi esistono diverse tecniche di approccio al calcolo dell’Health Index, quella che viene presentata è una tecnica del tutto innovativa, ossia sviluppare una rete neurale artificiale (Artificial Neural Network, ANN) in grado di prevedere lo stato del trasformatore basandosi su misure effettuate sullo stesso. Dunque, verranno presentante le basi per lo sviluppo di una rete neurale, partendo dall’analisi e formattazione dei dati, fino alla fase di ottimizzazione delle prestazioni. Infine, si attraverseranno tutte le fasi intermedie di realizzazione del progetto da cui l’elaborato prende il titolo; osservando l’evoluzione di una rete neurale che si trasforma da un programma scritto in ambiente Python a una applicazione pronta all’uso per gli operatori durante le operazioni di diagnostica.

Machine Learning algorithm on seat comfort level measurement, mapping pressure values with body regions

The comfort level of the seat has a major effect on the usage of a vehicle; thus, car manufacturers have been working on elevating car seat comfort as much as possible. However, still, the testing and evaluation of comfort are done using exhaustive trial and error testing and evaluation of data. In this thesis, we resort to machine learning and Artificial Neural Networks (ANN) to develop a fully automated approach. Even though this approach has its advantages in minimizing time and using a large set of data, it takes away the degree of freedom of the engineer on making decisions. The focus of this study is on filling the gap in a two-step comfort level evaluation which used pressure mapping with body regions to evaluate the average pressure supported by specific body parts and the Self-Assessment Exam (SAE) questions on evaluation of the person’s interest. This study has created a machine learning algorithm that works on giving a degree of freedom to the engineer in making a decision when mapping pressure values with body regions using ANN. The mapping is done with 92% accuracy and with the help of a Graphical User Interface (GUI) that facilitates the process during the testing time of comfort level evaluation of the car seat, which decreases the duration of the test analysis from days to hours.

Application of active and transfer learning to wound image segmentation

Wound management is a fundamental task in standard clinical practice. Automated solutions already exist for humans, but there is a lack of applications on wound management for pets. The importance of a precise and efficient wound assessment is helpful to improve diagnosis and to increase the effectiveness of treatment plans for the chronic wounds. The goal of the research was to propose an automated pipeline capable of segmenting natural light-reflected wound images of animals. Two datasets composed by light-reflected images were used in this work: Deepskin dataset, 1564 human wound images obtained during routine dermatological exams, with 145 manual annotated images; Petwound dataset, a set of 290 wound photos of dogs and cats with 0 annotated images. Two implementations of U-Net Convolutioal Neural Network model were proposed for the automated segmentation. Active Semi-Supervised Learning techniques were applied for human-wound images to perform segmentation from 10% of annotated images. Then the same models were trained, via Transfer Learning, adopting an Active Semi- upervised Learning to unlabelled animal-wound images. The combination of the two training strategies proved their effectiveness in generating large amounts of annotated samples (94% of Deepskin, 80% of PetWound) with the minimal human intervention. The correctness of automated segmentation were evaluated by clinical experts at each round of training thus we can assert that the results obtained in this thesis stands as a reliable solution to perform a correct wound image segmentation. The use of Transfer Learning and Active Semi-Supervied Learning allows to minimize labelling effort from clinicians, even requiring no starting manual annotation at all. Moreover the performances of the model with limited number of parameters suggest the implementation of smartphone-based application to this topic, helping the future standardization of light-reflected images as acknowledge medical images.

Continual learning on nano drones

Miniaturized flying robotic platforms, called nano-drones, have the potential to revolutionize the autonomous robots industry sector thanks to their very small form factor. The nano-drones’ limited payload only allows for a sub-100mW microcontroller unit for the on-board computations. Therefore, traditional computer vision and control algorithms are too computationally expensive to be executed on board these palm-sized robots, and we are forced to rely on artificial intelligence to trade off accuracy in favor of lightweight pipelines for autonomous tasks. However, relying on deep learning exposes us to the problem of generalization since the deployment scenario of a convolutional neural network (CNN) is often composed by different visual cues and different features from those learned during training, leading to poor inference performances. Our objective is to develop and deploy and adaptation algorithm, based on the concept of latent replays, that would allow us to fine-tune a CNN to work in new and diverse deployment scenarios. To do so we start from an existing model for visual human pose estimation, called PULPFrontnet, which is used to identify the pose of a human subject in space through its 4 output variables, and we present the design of our novel adaptation algorithm, which features automatic data gathering and labeling and on-device deployment. We therefore showcase the ability of our algorithm to adapt PULP-Frontnet to new deployment scenarios, improving the R2 scores of the four network outputs, with respect to an unknown environment, from approximately [−0.2, 0.4, 0.0,−0.7] to [0.25, 0.45, 0.2, 0.1]. Finally we demonstrate how it is possible to fine-tune our neural network in real time (i.e., under 76 seconds), using the target parallel ultra-low power GAP 8 System-on-Chip on board the nano-drone, and we show how all adaptation operations can take place using less than 2mWh of energy, a small fraction of the available battery power.

Structural Connectivity Of The Default Mode Network And Cognition In Alzheimer׳s Disease.

Disconnectivity between the Default Mode Network (DMN) nodes can cause clinical symptoms and cognitive deficits in Alzheimer׳s disease (AD). We aimed to examine the structural connectivity between DMN nodes, to verify the extent in which white matter disconnection affects cognitive performance. MRI data of 76 subjects (25 mild AD, 21 amnestic Mild Cognitive Impairment subjects and 30 controls) were acquired on a 3.0T scanner. ExploreDTI software (fractional Anisotropy threshold=0.25 and the angular threshold=60°) calculated axial, radial, and mean diffusivities, fractional anisotropy and streamline count. AD patients showed lower fractional anisotropy (P=0.01) and streamline count (P=0.029), and higher radial diffusivity (P=0.014) than controls in the cingulum. After correction for white matter atrophy, only fractional anisotropy and radial diffusivity remained significantly lower in AD compared to controls (P=0.003 and P=0.05). In the parahippocampal bundle, AD patients had lower mean and radial diffusivities (P=0.048 and P=0.013) compared to controls, from which only radial diffusivity survived for white matter adjustment (P=0.05). Regression models revealed that cognitive performance is also accounted for by white matter microstructural values. Structural connectivity within the DMN is important to the execution of high-complexity tasks, probably due to its relevant role in the integration of the network.