In the past two decades, autonomous driving has been catalyzed into reality by the growing capabilities of machine learning. This paradigm shift possesses significant potential to transform the future of mobility and reshape our society as a whole. With the recent advances in perception, planning, and control capabilities, autonomous driving technologies are being rolled out for public trials, yet we remain far from being able to rigorously ensure the resilient operations of these systems across the long-tailed nature of the driving environment. Given the limitations of real-world testing, autonomous vehicle simulation stands as the critical component in exploring the edge of autonomous driving capabilities, developing the robust behaviors required for successful real-world operation, and enabling the extraction of hidden risks from these complex systems prior to deployment. This paper presents the current state-of-the-art simulation frameworks and methodologies used in the development of autonomous driving systems, with a focus on outlining how simulation is used to build the resiliency required for real-world operation and the methods developed to bridge the gap between simulation and reality. A synthesis of the key challenges surrounding autonomous driving simulation is presented, specifically highlighting the opportunities to further advance the ability to continuously learn in simulation and effectively transfer the learning into the real-world - enabling autonomous vehicles to exit the guardrails of simulation and deliver robust and resilient operations at scale.

The recent advances in reinforcement learning have led to effective methods able to obtain above human-level performances in very complex environments. However, once solved, these environments become less valuable, and new challenges with different or more complex scenarios are needed to support research advances. This work presents DIAMBRA Arena, a new platform for reinforcement learning research and experimentation, featuring a collection of high-quality environments exposing a Python API fully compliant with OpenAI Gym standard. They are episodic tasks with discrete actions and observations composed by raw pixels plus additional numerical values, all supporting both single player and two players mode, allowing to work on standard reinforcement learning, competitive multi-agent, human-agent competition, self-play, human-in-the-loop training and imitation learning. Software capabilities are demonstrated by successfully training multiple deep reinforcement learning agents with proximal policy optimization obtaining human-like behavior. Results confirm the utility of DIAMBRA Arena as a reinforcement learning research tool, providing environments designed to study some of the most challenging topics in the field.

We study the problem of policy optimization (PO) with linear temporal logic (LTL) constraints. The language of LTL allows flexible description of tasks that may be unnatural to encode as a scalar cost function. We consider LTL-constrained PO as a systematic framework, decoupling task specification from policy selection, and as an alternative to the standard of cost shaping. With access to a generative model, we develop a model-based approach that enjoys a sample complexity analysis for guaranteeing both task satisfaction and cost optimality (through a reduction to a reachability problem). Empirically, our algorithm can achieve strong performance even in low-sample regimes.

There is a recent proliferation of research on the integration of machine learning and optimization. One expansive area within this research stream is predictive-model embedded optimization, which proposes the use of pre-trained predictive models as surrogates for uncertain or highly complex objective functions. In this setting, features of the predictive models become decision variables in the optimization problem. Despite a recent surge in publications in this area, only a few papers note the importance of incorporating trust region considerations in this decision-making pipeline, i.e., enforcing solutions to be similar to the data used to train the predictive models. Without such constraints, the evaluation of the predictive model at solutions obtained from optimization cannot be trusted and the practicality of the solutions may be unreasonable. In this paper, we provide an overview of the approaches appearing in the literature to construct a trust region, and propose three alternative approaches. Our numerical evaluation highlights that trust-region constraints learned through isolation forests, one of the newly proposed approaches, outperform all previously suggested approaches, both in terms of solution quality and computational time.

Despite the impressive performance of Artificial Intelligence (AI) systems, their robustness remains elusive and constitutes a key issue that impedes large-scale adoption. Robustness has been studied in many domains of AI, yet with different interpretations across domains and contexts. In this work, we systematically survey the recent progress to provide a reconciled terminology of concepts around AI robustness. We introduce three taxonomies to organize and describe the literature both from a fundamental and applied point of view: 1) robustness by methods and approaches in different phases of the machine learning pipeline; 2) robustness for specific model architectures, tasks, and systems; and in addition, 3) robustness assessment methodologies and insights, particularly the trade-offs with other trustworthiness properties. Finally, we identify and discuss research gaps and opportunities and give an outlook on the field. We highlight the central role of humans in evaluating and enhancing AI robustness, considering the necessary knowledge humans can provide, and discuss the need for better understanding practices and developing supportive tools in the future.

Since the 2004 DARPA Grand Challenge, the autonomous driving technology has witnessed nearly two decades of rapid development. Particularly, in recent years, with the application of new sensors and deep learning technologies extending to the autonomous field, the development of autonomous driving technology has continued to make breakthroughs. Thus, many carmakers and high-tech giants dedicated to research and system development of autonomous driving. However, as the foundation of autonomous driving, the deep learning technology faces many new security risks. The academic community has proposed deep learning countermeasures against the adversarial examples and AI backdoor, and has introduced them into the autonomous driving field for verification. Deep learning security matters to autonomous driving system security, and then matters to personal safety, which is an issue that deserves attention and research.This paper provides an summary of the concepts, developments and recent research in deep learning security technologies in autonomous driving. Firstly, we briefly introduce the deep learning framework and pipeline in the autonomous driving system, which mainly include the deep learning technologies and algorithms commonly used in this field. Moreover, we focus on the potential security threats of the deep learning based autonomous driving system in each functional layer in turn. We reviews the development of deep learning attack technologies to autonomous driving, investigates the State-of-the-Art algorithms, and reveals the potential risks. At last, we provides an outlook on deep learning security in the autonomous driving field and proposes recommendations for building a safe and trustworthy autonomous driving system.

In the past decade, the number of computer-aided-diagnosis (CAD) studies via Machine Learning (ML) and deep learning (DL) algorithms has grown exponentially and has seen an incredible spike in their applications, especially in the biomedical field [1, 2, 3, 4, 5, 6]. Their use in cancer detection is numerous and allows for rapid diagnosis of different cancer types. The world has seen an impressive increase in cancer cases. Not only the number has continuously grown; but with around 9 million cancer deaths in 2017 worldwide, and 2 million new cases and 600 thousand cancer deaths in 2021 in the United States alone [7], the need for appropriate tools of detection and diagnosis is becoming more and more pressing both for accuracy and rapidity [8]. According to the American Cancer Society, the chances of survival over 5 years for an individual increase by 90% when cancers are detected early. Similarly, screening for breast cancers has resulted in a lower death risk of 20-40% [9, 10]. However, lung cancers are detected in their later stages in 70 % of the cases [11]. In addition to decreasing chances of survival, late cancer detection leads to potential outbreaks of cancerous cells in other parts of the body leading to metastasis which needs to be prevented at all costs and provide a challenge for machine learning techniques. Amongst all cancers, the tracheal, bronchus, and lung cancers are the most prevalent ones with a little under 2 million deaths, closely followed by colon, stomach, and liver cancers (digestive tract cancers) with around 800 thousand deaths for the year of 2017.

Despite the remarkable technological advances that fill our lives today, the ways we work with the metals that underlie these developments haven't changed significantly in thousands of years. This is true of everything from the metal rods, tubes, and cubes that provide cars and trucks with their shape, strength, and fuel economy, to wires that move electrical energy in everything from motors to undersea cables. But things are changing rapidly: The materials manufacturing industry is using new and innovative technologies, processes, and methods to improve existing products and create new ones. Pacific Northwest National Laboratory (PNNL) is a leader in this space, known as advanced manufacturing. For example, scientists working in PNNL's Mathematics for Artificial Reasoning in Science initiative are pioneering approaches in the branch of artificial intelligence known as machine learning to design and train computer software programs that guide the development of new manufacturing processes.

Startups often represent newly established business models associated with disruptive innovation and high scalability. They are commonly regarded as powerful engines for economic and social development. Meanwhile, startups are heavily constrained by many factors such as limited financial funding and human resources. Therefore the chance for a startup to eventually succeed is as rare as ``spotting a unicorn in the wild''. Venture Capital (VC) strives to identify and invest in unicorn startups during their early stages, hoping to gain a high return. To avoid entirely relying on human domain expertise and intuition, investors usually employ data-driven approaches to forecast the success probability of startups. Over the past two decades, the industry has gone through a paradigm shift moving from conventional statistical approaches towards becoming machine-learning (ML) based. Notably, the rapid growth of data volume and variety is quickly ushering in deep learning (DL), a subset of ML, as a potentially superior approach in terms capacity and expressivity. In this work, we carry out a literature review and synthesis on DL-based approaches, covering the entire DL life cycle. The objective is a) to obtain a thorough and in-depth understanding of the methodologies for startup evaluation using DL, and b) to distil valuable and actionable learning for practitioners. To the best of our knowledge, our work is the first of this kind.

Machine learning (ML) formalizes the problem of getting computers to learn from experience as optimization of performance according to some metric(s) on a set of data examples. This is in contrast to requiring behaviour specified in advance (e.g. by hard-coded rules). Formalization of this problem has enabled great progress in many applications with large real-world impact, including translation, speech recognition, self-driving cars, and drug discovery. But practical instantiations of this formalism make many assumptions - for example, that data are i.i.d.: independent and identically distributed - whose soundness is seldom investigated. And in making great progress in such a short time, the field has developed many norms and ad-hoc standards, focused on a relatively small range of problem settings. As applications of ML, particularly in artificial intelligence (AI) systems, become more pervasive in the real world, we need to critically examine these assumptions, norms, and problem settings, as well as the methods that have become de-facto standards. There is much we still do not understand about how and why deep networks trained with stochastic gradient descent are able to generalize as well as they do, why they fail when they do, and how they will perform on out-of-distribution data. In this thesis I cover some of my work towards better understanding deep net generalization, identify several ways assumptions and problem settings fail to generalize to the real world, and propose ways to address those failures in practice.