In the age of artificial intelligence, Machine Learning (ML) has become a pervasive force, impacting countless aspects of our lives. As ML’s influence expands, concerns about its reliability and trustworthiness have intensified, with security and robustness emerging as significant challenges. For instance, it has been demonstrated that slight perturbations to a stop sign can cause ML classifiers to misidentify it as a speed limit sign, raising concerns about whether ML algorithms are suitable for real-world deployments. To tackle these issues, Responsible Machine Learning (Responsible ML) has emerged with a clear mission: to develop secure and robust ML algorithms. This dissertation aims to develop Responsible Machine Learning algorithms under real-world constraints. Specifically, recognizing the role of adversarial attacks in exposing security vulnerabilities and robustifying the ML methods, it lays down the foundation of Responsible ML by outlining a novel taxonomy of adversarial attacks within real-world settings, categorizing them into black-box target-specific, and target-agnostic attacks. Subsequently, it proposes potent adversarial attacks in each category, aiming to obtain effectiveness and efficiency. Transcending conventional boundaries, it then introduces the notion of causality into Responsible ML (a.k.a., Causal Responsible ML), presenting the causal adversarial attack. This represents the first principled framework to explain the transferability of adversarial attacks to unknown models by identifying their common source of vulnerabilities, thereby exposing the pinnacle of threat and vulnerability: conducting successful attacks on any model with no prior knowledge. Finally, acknowledging the surge of Generative AI, this dissertation explores Responsible ML for Generative AI. It introduces a novel adversarial attack that unveils their adversarial vulnerabilities and devises a strong defense mechanism to bolster the models’ robustness against potential attacks.

In the rapidly evolving field of computer vision, propelled by advancements in deeplearning, the integration of hardware-software co-design has become crucial to overcome the limitations of traditional imaging systems. This dissertation explores the integration of hardware-software co-design in computational imaging, particularly in light transport acquisition and Non-Line-of-Sight (NLOS) imaging. By leveraging projector-camera systems and computational techniques, this thesis address critical challenges in imaging complex environments, such as adverse weather conditions, low-light scenarios, and the imaging of reflective or transparent objects. The first contribution in this thesis is the theory, design, and implementation of a slope disparity gating system, which is a vertically aligned configuration of a synchronized raster scanning projector and rolling-shutter camera, facilitating selective imaging through disparity-based triangulation. This system introduces a novel, hardware-oriented approach to selective imaging, circumventing the limitations of post-capture processing. The second contribution of this thesis is the realization of two innovative approaches for spotlight optimization to improve localization and tracking for NLOS imaging. The first approach utilizes radiosity-based optimization to improve 3D localization and object identification for small-scale laboratory settings. The second approach introduces a learningbased illumination network along with a differentiable renderer and NLOS estimation network to optimize human 2D localization and activity recognition. This approach is validated on a large, room-scale scene with complex line-of-sight geometries and occluders. The third contribution of this thesis is an attention-based neural network for passive NLOS settings where there is no controllable illumination. The thesis demonstrates realtime, dynamic NLOS human tracking where the camera is moving on a mobile robotic platform. In addition, this thesis contains an appendix featuring temporally consistent relighting for portrait videos with applications in computer graphics and vision.

Millimeter-wave (mmWave) and sub-terahertz (sub-THz) systems aim to utilize the large bandwidth available at these frequencies. This has the potential to enable several future applications that require high data rates, such as autonomous vehicles and digital twins. These systems, however, have several challenges that need to be addressed to realize their gains in practice. First, they need to deploy large antenna arrays and use narrow beams to guarantee sufficient receive power. Adjusting the narrow beams of the large antenna arrays incurs massive beam training overhead. Second, the sensitivity to blockages is a key challenge for mmWave and THz networks. Since these networks mainly rely on line-of-sight (LOS) links, sudden link blockages highly threaten the reliability of the networks. Further, when the LOS link is blocked, the network typically needs to hand off the user to another LOS basestation, which may incur critical time latency, especially if a search over a large codebook of narrow beams is needed. A promising way to tackle both these challenges lies in leveraging additional side information such as visual, LiDAR, radar, and position data. These sensors provide rich information about the wireless environment, which can be utilized for fast beam and blockage prediction. This dissertation presents a machine-learning framework for sensing-aided beam and blockage prediction. In particular, for beam prediction, this work proposes to utilize visual and positional data to predict the optimal beam indices. For the first time, this work investigates the sensing-aided beam prediction task in a real-world vehicle-to-infrastructure and drone communication scenario. Similarly, for blockage prediction, this dissertation proposes a multi-modal wireless communication solution that utilizes bimodal machine learning to perform proactive blockage prediction and user hand-off. Evaluations on both real-world and synthetic datasets illustrate the promising performance of the proposed solutions and highlight their potential for next-generation communication and sensing systems.

This thesis presents robust and novel solutions using knowledge distillation with geometric approaches and multimodal data that can address the current challenges in deep learning, providing a comprehensive understanding of the learning process involved in knowledge distillation. Deep learning has attained significant success in various applications, such as health and wellness promotion, smart homes, and intelligent surveillance. In general, stacking more layers or increasing the number of trainable parameters causes deep networks to exhibit improved performance. However, this causes the model to become large, resulting in an additional need for computing and power resources for training, storage, and deployment. These are the core challenges in incorporating such models into small devices with limited power and computational resources. In this thesis, robust solutions aimed at addressing the aforementioned challenges are presented. These proposed methodologies and algorithmic contributions enhance the performance and efficiency of deep learning models. The thesis encompasses a comprehensive exploration of knowledge distillation, an approach that holds promise for creating compact models from high-capacity ones, while preserving their performance. This exploration covers diverse datasets, including both time series and image data, shedding light on the pivotal role of augmentation methods in knowledge distillation. The effects of these methods are rigorously examined through empirical experiments. Furthermore, the study within this thesis delves into the efficient utilization of features derived from two different teacher models, each trained on dissimilar data representations, including time-series and image data. Through these investigations, I present novel approaches to knowledge distillation, leveraging geometric techniques for the analysis of multimodal data. These solutions not only address real-world challenges but also offer valuable insights and recommendations for modeling in new applications.

Generative models are deep neural network-based models trained to learn the underlying distribution of a dataset. Once trained, these models can be used to sample novel data points from this distribution. Their impressive capabilities have been manifested in various generative tasks, encompassing areas like image-to-image translation, style transfer, image editing, and more. One notable application of generative models is data augmentation, aimed at expanding and diversifying the training dataset to augment the performance of deep learning models for a downstream task. Generative models can be used to create new samples similar to the original data but with different variations and properties that are difficult to capture with traditional data augmentation techniques. However, the quality, diversity, and controllability of the shape and structure of the generated samples from these models are often directly proportional to the size and diversity of the training dataset. A more extensive and diverse training dataset allows the generative model to capture overall structures present in the data and generate more diverse and realistic-looking samples. In this dissertation, I present innovative methods designed to enhance the robustness and controllability of generative models, drawing upon physics-based, probabilistic, and geometric techniques. These methods help improve the generalization and controllability of the generative model without necessarily relying on large training datasets. I enhance the robustness of generative models by integrating classical geometric moments for shape awareness and minimizing trainable parameters. Additionally, I employ non-parametric priors for the generative model's latent space through basic probability and optimization methods to improve the fidelity of interpolated images. I adopt a hybrid approach to address domain-specific challenges with limited data and controllability, combining physics-based rendering with generative models for more realistic results. These approaches are particularly relevant in industrial settings, where the training datasets are small and class imbalance is common. Through extensive experiments on various datasets, I demonstrate the effectiveness of the proposed methods over conventional approaches.

Quantum computing has the potential to revolutionize the signal-processing field by providing more efficient methods for analyzing signals. This thesis explores the application of quantum computing in signal analysis synthesis for compression applications. More specifically, the study focuses on two key approaches: quantum Fourier transform (QFT) and quantum linear prediction (QLP). The research is motivated by the potential advantages offered by quantum computing in massive signal processing tasks and presents novel quantum circuit designs for QFT, quantum autocorrelation, and QLP, enabling signal analysis synthesis using quantum algorithms. The two approaches are explained as follows. The Quantum Fourier transform (QFT) demonstrates the potential for improved speed in quantum computing compared to classical methods. This thesis focuses on quantum encoding of signals and designing quantum algorithms for signal analysis synthesis, and signal compression using QFTs. Comparative studies are conducted to evaluate quantum computations for Fourier transform applications, considering Signal-to-Noise-Ratio results. The effects of qubit precision and quantum noise are also analyzed. The QFT algorithm is also developed in the J-DSP simulation environment, providing hands-on laboratory experiences for signal-processing students. User-friendly simulation programs on QFT-based signal analysis synthesis using peak picking, and perceptual selection using psychoacoustics in the J-DSP are developed. Further, this research is extended to analyze the autocorrelation of the signal using QFTs and develop a quantum linear prediction (QLP) algorithm for speech processing applications. QFTs and IQFTs are used to compute the quantum autocorrelation of the signal, and the HHL algorithm is modified and used to compute the solutions of the linear equations using quantum computing. The performance of the QLP algorithm is evaluated for system identification, spectral estimation, and speech analysis synthesis, and comparisons are performed for QLP and CLP results. The results demonstrate the following: effective quantum circuits for accurate QFT-based speech analysis synthesis, evaluation of performance with quantum noise, design of accurate quantum autocorrelation, and development of a modified HHL algorithm for efficient QLP. Overall, this thesis contributes to the research on quantum computing for signal processing applications and provides a foundation for further exploration of quantum algorithms for signal analysis synthesis.

This thesis encompasses a comprehensive research effort dedicated to overcoming the critical bottlenecks that hinder the current generation of neural networks, thereby significantly advancing their reliability and performance. Deep neural networks, with their millions of parameters, suffer from over-parameterization and lack of constraints, leading to limited generalization capabilities. In other words, the complex architecture and millions of parameters present challenges in finding the right balance between capturing useful patterns and avoiding noise in the data. To address these issues, this thesis explores novel solutions based on knowledge distillation, enabling the learning of robust representations. Leveraging the capabilities of large-scale networks, effective learning strategies are developed. Moreover, the limitations of dependency on external networks in the distillation process, which often require large-scale models, are effectively overcome by proposing a self-distillation strategy. The proposed approach empowers the model to generate high-level knowledge within a single network, pushing the boundaries of knowledge distillation. The effectiveness of the proposed method is not only demonstrated across diverse applications, including image classification, object detection, and semantic segmentation but also explored in practical considerations such as handling data scarcity and assessing the transferability of the model to other learning tasks. Another major obstacle hindering the development of reliable and robust models lies in their black-box nature, impeding clear insights into the contributions toward the final predictions and yielding uninterpretable feature representations. To address this challenge, this thesis introduces techniques that incorporate simple yet powerful deep constraints rooted in Riemannian geometry. These constraints confer geometric qualities upon the latent representation, thereby fostering a more interpretable and insightful representation. In addition to its primary focus on general tasks like image classification and activity recognition, this strategy offers significant benefits in real-world applications where data scarcity is prevalent. Moreover, its robustness in feature removal showcases its potential for edge applications. By successfully tackling these challenges, this research contributes to advancing the field of machine learning and provides a foundation for building more reliable and robust systems across various application domains.

This project explores the potential for the accurate prediction of basketball shooting posture with machine learning (ML) prediction algorithms, using the data collected by an Internet of Things (IoT) based motion capture system. Specifically, this question is addressed in the research - Can I develop an ML model to generalize a decent basketball shot pattern? - by introducing a supervised learning paradigm, where the ML method takes acceleration attributes to predict the basketball shot efficiency. The solution presented in this study considers motion capture devices configuration on the right upper limb with a sole motion sensor made by BNO080 and ESP32 attached on the right wrist, right forearm, and right shoulder, respectively, By observing the rate of speed changing in the shooting movement and comparing their performance, ML models that apply K-Nearest Neighbor, and Decision Tree algorithm, conclude the best range of acceleration that different spots on the arm should implement.

Simultaneous localization and mapping (SLAM) has traditionally relied on low-level geometric or optical features. However, these features-based SLAM methods often struggle with feature-less or repetitive scenes. Additionally, low-level features may not provide sufficient information for robot navigation and manipulation, leaving robots without a complete understanding of the 3D spatial world. Advanced information is necessary to address these limitations. Fortunately, recent developments in learning-based 3D reconstruction allow robots to not only detect semantic meanings, but also recognize the 3D structure of objects from a few images. By combining this 3D structural information, SLAM can be improved from a low-level approach to a structure-aware approach. This work propose a novel approach for multi-view 3D reconstruction using recurrent transformer. This approach allows robots to accumulate information from multiple views and encode them into a compact latent space. The resulting latent representations are then decoded to produce 3D structural landmarks, which can be used to improve robot localization and mapping.

In the era of data explosion, massive data is generated from various sources at an unprecedented speed. The ever-growing amount of data reveals enormous opportunities for developing novel data-driven solutions to unsolved problems. In recent years, benefiting from numerous public datasets and advances in deep learning, data-driven approaches in the computer vision domain have demonstrated superior performance with high adaptability on various data and tasks. Meanwhile, signal processing has long been dominated by techniques derived from rigorous mathematical models built upon prior knowledge of signals. Due to the lack of adaptability to real data and applications, model-based methods often suffer from performance degradation and engineering difficulties. In this dissertation, multiple signal processing problems are studied from vision-inspired data representation and learning perspectives to address the major limitation on adaptability. Corresponding data-driven solutions are proposed to achieve significantly improved performance over conventional solutions. Specifically, in the compressive sensing domain, an open-source image compressive sensing toolbox and benchmark to standardize the implementation and evaluation of reconstruction methods are first proposed. Then a plug-and-play compression ratio adapter is proposed to enable the adaptability of end-to-end data-driven reconstruction methods to variable compression ratios. Lastly, the problem of transfer learning from images to bioelectric signals is experimentally studied to demonstrate the improved performance of data-driven reconstruction. In the image subsampling domain, task-adaptive data-driven image subsampling is studied to reduce data redundancy and retain information of interest simultaneously. In the semiconductor analysis domain, the data-driven automatic error detection problem is studied in the context of integrated circuit segmentation for the first time. In the light detection and ranging(LiDAR) camera calibration domain, the calibration accuracy degradation problem in low-resolution LiDAR scenarios is addressed with data-driven techniques.