Accurate reference tracking is essential in control tasks in a variety of applications, and, in repetitive systems, model-based Iterative Learning Control (ILC) is a standard solution that is underpinned by formal stability and convergence guarantees. To overcome the limitations of model-based ILC design, data-driven ILC methods have been introduced which lead to a patchwork of several, very different approaches that do not preserve the modularity and theoretical guarantees of the many, well-established model-based ILC methods. We, for the first time, propose a unifying framework for data-driven ILC that preserves the modularity and formally proven properties of model-based ILC. Specifically, we propose Iterative Model Learning (IML) and Dual Iterative Learning Control (DILC). The IML framework enables iterative learning of unknown dynamics in repetitive systems using input/output trajectory pairs. We formally prove the duality between IML and ILC, i.e., an IML system is equivalent to an ILC system with a trial-varying reference and trial-varying but known dynamics. We further provide generic conditions to guarantee the convergence of the model and prove the duality of monotonic convergence in ILC and IML. This duality allows arbitrary, established, model-based ILC methods to be effectively applied within the IML framework to learn models of unknown dynamics. Unlike existing data-driven ILC approaches which only combine specific model learning schemes with specific model-based ILC approaches, the proposed DILC framework combines IML with model-based ILC such that established, model-based ILC methods can be arbitrarily combined for simultaneous model and control learning with a formally proven separation principle. The proposed DILC is a unified framework for data-driven ILC that not only relieves model-based ILC of a priori model requirement but also preserves the modularity and theoretical guarantees of model-based ILC. Both IML and DILC are validated by extensive simulations and real-world experiments.

Quantum networks connect devices to facilitate the transmission of quantum bits (qubits), harnessing quantum mechanics for groundbreaking telecommunications applications. The quantum state exchange rely on quantum entangled particles which have to be sent from a Quantum Base Station (QBS) to Quantum Nodes (QNs) through an optical channel. Free Space Optical (FSO) links promise flexibility and cost advantages over the classical fiber-based infrastructure. However, the absence of Line of Sight (LoS) between the QBS and QNs may completely disrupt the communication. Therefore, this work focuses on an Intelligent Reflective Surface (IRS)-assisted Unmanned Aerial Vehicle (UAV)-aided FSO quantum network. An optimization problem is formulated to fairly maximize qubit reception at QNs by optimizing UAV hovering locations, constrained by fidelity and link-per-IRS limits. The intractability of the original problem is first tackled with a sigmoid-based approximations, subsequently split into three sub-problems, sequentially solved applying the Successive Convex Approximation (SCA) technique. Numerical results (i) validate the approximation accuracy, and (ii) demonstrates the effectiveness of the proposal against two baseline approaches, showing an order ofmagnitude increase in terms of fairness.

This work presents a novel signal design based on time-domain interleaving (TDI) of orthogonal frequency division multiplexing (OFDM). Unlike conventional OFDM-TDI that introduces a one-dimensional frequency diversity, the proposed design introduces a two-dimensional frequency diversity leading to a superior symbol error rate (SER) performance for various channel conditions. The performance of the proposed signal design is evaluated using zero-forcing (ZF) and minimum-mean-square-error (MMSE) equalizers where the signal-to-noise ratio (SNR) and signal-to-interference-plus-noise ratio (SINR) are derived and used to evaluate the SER, outage probability (OP), and diversity order. The analytical SER results obtained, corroborated by Monte Carlo simulation, demonstrate that the proposed signal model with MMSE equalizer achieves ∼40 dB gain over OFDM and ∼7 dB over conventional TDI in Rayleigh fading channels, and it is only 2 dB from the Gaussian channel SER in Rician channels. The proposed scheme also outperforms other state-of-the-art techniques such as interleave frequency division multiplexing (IFDM) and orthogonal time frequency space (OTFS). The proposed system architecture generally follows the conventional OFDM system model, which emphasizes its compatibility and low complexity.

Electric vehicles (EVs) are playing a pivotal role in transportation systems to conform to the rising exigencies for enhanced performance with safety and hindered environmental impact. Thus, to improve the efficiency and downsize the maintenance cost of EVs, an early fault diagnosis (FD) framework is crucial. Bearing and stator winding faults, which account for approximately 78% of induction machine (IM) incipient faults in EVs, remain rather elusive for conventional sensors to accurately classify them and their incipient values. To this end, this paper presents a novel Attention-enhanced Autoencoder-Gated Recurrent Unit (AAGRU) model that improves the accuracy and efficiency of fault analysis in IMs. Moreover, since bearing faults are usually captured through vibration sensors, which are expensive and require direct coupling with the IM, a hybrid signal re-constructor is devised based on merely stator current signals. The proposed model leverages Empirical Mode Decomposition (EMD), Fast Fourier Transform (FFT), and Discrete Wavelet Decomposition (DWT) to process the electric current data, which are then used by the AAGRU model to identify, detect, and classify the fault patterns. Experimental results demonstrate that the proposed model offers a 6%-20% improvement in fault detection and an 8%-28% improvement in inter-turn shortcircuit fault severity classification relative to different shallow and deep-based benchmark models. The proposed model was also tested on different load conditions to ensure its applicability in practice.

Orthogonal Time Frequency and Space (OTFS) modulation-based system for Integrated Sensing and Communication (ISAC) in vehicular scenarios is a promising technology and is expected to provide reliable communication in high-speed environments. A vehicle-to-infrastructure-based OTFS multi-static system with single or multiple transmitters is considered in this work. The cell towers communicate to the intended target vehicle(s) and simultaneously estimate the location, velocity, and bistatic radar cross-section of other targets in the scene. Bringing in a multi-static framework gives the advantage of exploiting spatial diversity for better target detection. Pilot sequences are chosen to ensure both spectral efficiency and accurate detection and parameter estimation by using the signals from multiple transmitters. However, in the case of multiple targets, associating measurements to targets is critical in multi-static systems. The prime problem of multi-target data association is solved by a joint association and localization approach using a sparse recovery framework. Simulations are performed with real-time system and signal parameters. Performance of the algorithm for various OTFS Delay-Doppler block-size configurations and multiple targets of different bistatic radar cross-sections is demonstrated. The performance and complexity comparison with other existing data association approaches are furnished. The results highlight better accuracy with lesser complexity, thus aiding in the practical implementation of an ISAC system. Furthermore, the extension of our algorithm to accommodate fractional bistatic delay and Doppler measurements is also described so that enhanced performance can be achieved even with only modest bandwidth requirements.

A document by Nicolò La Porta. Click on the document to view its contents.

Ransomware, a malicious software type leveraging cryptography, has evolved into a significant global cybersecurity threat. This paper explores the use of symmetric, asymmetric, and hybrid encryption techniques in ransomware attacks and its impacts. Symmetric cryptography is favored for its speed, while asymmetric systems provide enhanced key security. Hybrid systems, combining the strengths of both, offer the most effective encryption method for ransomware operations. Through analysis of notable ransomware like Jigsaw, Archiveus, and WannaCry, this paper highlights their cryptographic methods, impact, and operational strategies. The study underscores the importance of robust countermeasures, including updated cybersecurity protocols and awareness, to mitigate ransomware risks. Additionally, the paper evaluates the role of evolution in ransomware attacks, including the importance of cryptocurrencies enabling anonymous ransom payments. The findings emphasize the need for comprehensive strategies that incorporate both technological advancements and human-centered approaches to cybersecurity. By addressing these challenges, individuals and organizations can reduce the frequency and severity of ransomware attacks.

In this paper, we propose a novel method for compressing and optimizing deep neural networks through channel pruning based on spectral norm. As deep learning models grow in complexity, the demand for efficient deployment, particularly in resource-constrained environments, has increased. Traditional pruning methods often rely on magnitude-based criteria, which can overlook the model's stability and generalization capabilities. In contrast, our approach leverages the spectral norm, the largest singular value of a layer's weight matrix, as a more principled criterion for pruning decisions. The spectral norm provides a measure of the network's sensitivity to input perturbations, offering a theoretically grounded method for identifying and removing less critical channels without significantly impacting model performance. We evaluate our method on CIFAR-10 with VGG-16 and ImageNet with ResNet-50, achieving substantial reductions in both model size and computational cost. Our approach leads to a decrease of over 80% in parameters and a 50% reduction in floating-point operations (FLOPs), while maintaining or even improving accuracy compared to traditional pruning techniques. These results demonstrate that spectral norm-based pruning is a robust and efficient method for deep network compression, opening avenues for further research into stability-aware optimization techniques.

Scene boundary detection is crucial in applications like scene segmentation and video skimming, but the presence of ad clips makes it difficult. There are some existing methods based on visual, audiovisual and audio-only features. The existing audio-only feature-based method depends on short silences between program and ad or between ads. Still, short silences can also be present inside the program, affecting performance. So we proposed a detection method for ad shots using MFCC features. MFCC is a basic speech feature based on short-term spectral. We used the average of MFCC for a shot and determined the threshold for detection purposes. We used four episodes of The Big Bang Theory as our dataset. Our method is the first method that uses MFCC features for ad detection and explores the relation between video contents and MFCC features.

This paper extends the design of an autonomous cyber defence (ACD) agent to monitor and actuate within a protected core network segment. The goal is to take advantage of recent developments in AI models to define a hybrid architecture that combines deep reinforcement learning (DRL), large language models (LLMs), and rule-based models. The motivation comes from the fact that modern network segments within colored clouds are using software-defined controllers with the means to host ACD agents and other cybersecurity tools implementing hybrid AI models. For example, our ACD agent uses a DRL model and the chatbot uses an LLM to create an interface with human cybersecurity experts. The ACD agent was evaluated against two red agent strategies in a gym environment using a set of actions to defend services in the network (monitor, analyse, decoy, remove, and restore). Our chatbot was developed using retrieval augmented generation and a prompting agent to augment a pre-trained LLM with data from cybersecurity knowledge graphs. We performed a comparative analysis between a baseline implementation and our chatbot using generation/retrieval metrics. The results suggest that both ACD agent and chatbot can potentially enhance the defence of critical networks connected to untrusted infrastructure. Published version (DOI).

The inception of spatial transcriptomics has allowed improved comprehension of tissue architectures and the disentanglement of complex underlying biological, physiological, and pathological processes through their positional contexts. Recently, these contexts, and by extension the field, have seen much promise and elucidation with the application of graph learning approaches. In particular, neural operators have risen in regards to learning the mapping between infinite-dimensional function spaces. With basic to deep neural network architectures being data-driven, i.e. dependent on quality data for prediction, neural operators provide robustness by offering generalization among different resolutions despite low quality data. Graph neural operators are a variant that utilize graph networks to learn this mapping between function spaces. The aim of this research is to identify robust machine learning architectures that integrate spatial information to predict tissue types. Under this notion, we propose a study to validate the efficacy of applying neural operators towards classification of brain regions in mouse brain tissue samples as a proof of concept towards our purpose and compare it against various state of the art graph neural network approaches. We were able to achieve an F1 score of nearly 72% for the graph neural operator approach which outperformed all baseline and other graph network approaches within the scope of supervised learning.

Deep Neural Networks (DNNs) have gained considerable traction in recent years due to the unparalleled results they gathered. However, the cost behind training such sophisticated models is resource intensive, resulting in many to consider DNNs to be intellectual property (IP) to model owners. In this era of cloud computing, high-performance DNNs are often deployed all over the internet so that people can access them publicly. As such, DNN watermarking schemes, especially backdoor-based watermarks, have been actively developed in recent years to preserve proprietary rights. (Add that no backdoor watermark guarantee). Nonetheless, there lies much uncertainty on the robustness of existing backdoor watermark schemes, towards both adversarial attacks and unintended means such as fine-tuning neural network models. In this paper, we extensively evaluate the persistence of recent backdoor-based watermarks within neural networks in the scenario of fine-tuning, we propose/develop a novel data-driven idea to restore watermark after fine-tuning without exposing the trigger set. Our empirical results show that by solely introducing training data after fine-tuning, the watermark can be restored if model parameters do not shift dramatically during fine-tuning. Depending on the types of trigger samples used, trigger accuracy can be reinstated to up to 100%. Our study further explores how the restoration process works using loss landscape visualization, as well as the idea of introducing training data in fine-tuning stage to alleviate watermark vanishing.

A document by Praveer Dubey. Click on the document to view its contents.

Background: Document forgery poses a significant challenge to the integrity of sensitive documents and necessitates the development of advanced anti-forgery techniques. This study explored the integration of piezoelectric nanomaterials, specifically zinc oxide (ZnO) nanowires and barium titanate (BaTiO3) nanoparticles, into document fibers for enhanced forgery detection. Methods: This review examines the superior mechanical properties of ZnO nanowires and the high dielectric constant, piezoelectricity, and ferroelectricity of BaTiO3 nanoparticles, making them suitable for embedding within document fibers. Advanced scanning probe techniques, such as piezoresponse force microscopy (PFM), scanning Kelvin probe microscopy (SKPM), and electrostatic force microscopy (EFM), have been used to characterize the electrical properties and structural integrity of nanomaterials. Results: Embedding piezoelectric nanomaterials within document fibers enables the detection of tampering through changes in electrical properties, providing a non-invasive and highly sensitive forgery detection method. This study highlights the importance of optimizing the manufacturing processes, developing rigorous measurement protocols, and exploring alternative piezoelectric nanomaterials. Conclusions: The integration of piezoelectric nanomaterials into document fibers is a promising approach for addressing the persistent challenges of document forgery. Future research should focus on refining manufacturing techniques, improving measurement accuracy, and expanding the applications to various sensitive documents.

Hand neuroprostheses restore voluntary movement in people with paralysis through neuromodulation protocols. There are a variety of strategies to control hand neuroprostheses, which can be based on residual body movements or brain activity. There is no universally superior solution, rather the best approach may vary from patient to patient. Here, we propose a protocol based on an immersive virtual reality (VR) environment that simulates the use of a hand neuroprosthesis to allow patients to experience and familiarize themselves with various control schemes in clinically relevant tasks and choose the preferred one. We used our VR environment to compare two alternative control strategies over 5 days of training in four patients with C6 spinal cord injury: (a) control via the ipsilateral wrist, (b) control via the contralateral shoulder. We did not find a one-fits-all solution but rather a subject-specific preference that could not be predicted based only on a general clinical assessment. The main results were that the VR simulation allowed participants to experience the pros and cons of the proposed strategies and make an educated choice, and that there was a longitudinal improvement. This shows that our VR-based protocol is a useful tool for personalization and training of the control strategy of hand neuroprostheses, which could help to promote user comfort and thus acceptance.

This work presents the complete design process of a 3D fully-metallic reflectarray. The structure is composed of two unit cells with resonators in z direction that are phase-shifted 180 • and distributed in the x-y direction. Here will be shown how those unit cells have been designed and how we determine their distribution. Furthermore, a tolerance analysis is carried out with the aim of obtaining a prototype that is robust to manufacturing imperfections. Lastly, that protoype is measured to verify the results.

Human-Robot Interaction has been a topic of great interest from both science fiction and research point of view over many decades. Human-Robot Interaction is the study of interactions between humans and robots,which can be carried out through various means like:audio commands,visual intelligence and more. With the recent advancements in the field of computer vision, it has been possible to program computers to detect various objects and their features. For example,computer vision is being used for face recognition, color detection, and gesture recognition as well. The aim of this study is to harness the capability of computer vision in order to interact with a robotic hand system facilitating robotic telekinesis. The input to the system is the image frames taken from the web camera. The computer vision then detects various landmarks of the palm or the hand,thereby detecting the lengths between specific landmarks(therefore distance between hand and camera is critical).Then these lengths are fed to the micro-controller which further commands the robotic hand connected to the microcontroller to move accordingly.

Our research explores the integration of Artificial Intelligence (AI) within mirrorless cameras to automate color grading and editing processes, utilizing Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs). The project aims to harness AI's potential to analyze and learn from a photographer's historical stylistic choices, enabling personalized image enhancement directly in-camera. Current DSLR and mirrorless cameras already leverage AI for various functions such as autofocus, noise reduction, and image stabilization. Building on these applications, my project seeks to add a layer of creative automation by training models on a photographer's edited images, thereby recognizing and applying their unique stylistic traits to new photos. Our approach aims to streamline the editing workflow, offer inspiration, and maintain artistic individuality, addressing concerns over the potential homogenization of creative expression in photography. By embedding this technology directly into camera hardware, it provides an innovative tool for photographers to instantly visualize and apply their preferred aesthetics, challenging the traditional post-processing workflow. Our research not only advances the technical capabilities of camera systems but also explores the balance between technological innovation and the preservation of artistic integrity in this age of photography.