Artificial Intelligence (AI) is increasingly used in critical applications. Thus, the need for dependable AI systems is rapidly growing. In 2018, the European Commission appointed experts to a High-Level Expert Group on AI (AI-HLEG). AI-HLEG defined Trustworthy AI as 1) lawful, 2) ethical, and 3) robust and specified seven corresponding key requirements. To help development organizations, AI-HLEG recently published the Assessment List for Trustworthy AI (ALTAI). We present an illustrative case study from applying ALTAI to an ongoing development project of an Advanced Driver-Assistance System (ADAS) that relies on Machine Learning (ML). Our experience shows that ALTAI is largely applicable to ADAS development, but specific parts related to human agency and transparency can be disregarded. Moreover, bigger questions related to societal and environmental impact cannot be tackled by an ADAS supplier in isolation. We present how we plan to develop the ADAS to ensure ALTAI-compliance. Finally, we provide three recommendations for the next revision of ALTAI, i.e., life-cycle variants, domain-specific adaptations, and removed redundancy.

Artificial intelligence (AI) has been successful at solving numerous problems in machine perception. In radiology, AI systems are rapidly evolving and show progress in guiding treatment decisions, diagnosing, localizing disease on medical images, and improving radiologists' efficiency. A critical component to deploying AI in radiology is to gain confidence in a developed system's efficacy and safety. The current gold standard approach is to conduct an analytical validation of performance on a generalization dataset from one or more institutions, followed by a clinical validation study of the system's efficacy during deployment. Clinical validation studies are time-consuming, and best practices dictate limited re-use of analytical validation data, so it is ideal to know ahead of time if a system is likely to fail analytical or clinical validation. In this paper, we describe a series of sanity tests to identify when a system performs well on development data for the wrong reasons. We illustrate the sanity tests' value by designing a deep learning system to classify pancreatic cancer seen in computed tomography scans.

Deep learning models (with neural networks) have been widely used in challenging tasks such as computer-aided disease diagnosis based on medical images. Recent studies have shown deep diagnostic models may not be robust in the inference process and may pose severe security concerns in clinical practice. Among all the factors that make the model not robust, the most serious one is adversarial examples. The so-called "adversarial example" is a well-designed perturbation that is not easily perceived by humans but results in a false output of deep diagnostic models with high confidence. In this paper, we evaluate the robustness of deep diagnostic models by adversarial attack. Specifically, we have performed two types of adversarial attacks to three deep diagnostic models in both single-label and multi-label classification tasks, and found that these models are not reliable when attacked by adversarial example. We have further explored how adversarial examples attack the models, by analyzing their quantitative classification results, intermediate features, discriminability of features and correlation of estimated labels for both original/clean images and those adversarial ones. We have also designed two new defense methods to handle adversarial examples in deep diagnostic models, i.e., Multi-Perturbations Adversarial Training (MPAdvT) and Misclassification-Aware Adversarial Training (MAAdvT). The experimental results have shown that the use of defense methods can significantly improve the robustness of deep diagnostic models against adversarial attacks.

In this study, we explore the implications of integrating social distancing with emergency evacuation when a hurricane approaches a major city during the COVID-19 pandemic. Specifically, we compare DNN (Deep Neural Network)-based and non-DNN methods for generating evacuation strategies that minimize evacuation time while allowing for social distancing in rescue vehicles. A central question is whether a DNN-based method provides sufficient extra efficiency to accommodate social distancing, in a time-constrained evacuation operation. We describe the problem as a Capacitated Vehicle Routing Problem and solve it using one non-DNN solution (Sweep Algorithm) and one DNN-based solution (Deep Reinforcement Learning). DNN-based solution can provide decision-makers with more efficient routing than non-DNN solution. Although DNN-based solution can save considerable time in evacuation routing, it does not come close to compensating for the extra time required for social distancing and its advantage disappears as the vehicle capacity approaches the number of people per household.

Two popular types of machine translation (MT) are phrase-based and neural machine translation systems. Both of these types of systems are composed of multiple complex models or layers. Each of these models and layers learns different linguistic aspects of the source language. However, for some of these models and layers, it is not clear which linguistic phenomena are learned or how this information is learned. For phrase-based MT systems, it is often clear what information is learned by each model, and the question is rather how this information is learned, especially for its phrase reordering model. For neural machine translation systems, the situation is even more complex, since for many cases it is not exactly clear what information is learned and how it is learned. To shed light on what linguistic phenomena are captured by MT systems, we analyze the behavior of important models in both phrase-based and neural MT systems. We consider phrase reordering models from phrase-based MT systems to investigate which words from inside of a phrase have the biggest impact on defining the phrase reordering behavior. Additionally, to contribute to the interpretability of neural MT systems we study the behavior of the attention model, which is a key component in neural MT systems and the closest model in functionality to phrase reordering models in phrase-based systems. The attention model together with the encoder hidden state representations form the main components to encode source side linguistic information in neural MT. To this end, we also analyze the information captured in the encoder hidden state representations of a neural MT system. We investigate the extent to which syntactic and lexical-semantic information from the source side is captured by hidden state representations of different neural MT architectures.

Despite the success of convolutional neural networks (CNNs) in many computer vision and image analysis tasks, they remain vulnerable against so-called adversarial attacks: Small, crafted perturbations in the input images can lead to false predictions. A possible defense is to detect adversarial examples. In this work, we show how analysis in the Fourier domain of input images and feature maps can be used to distinguish benign test samples from adversarial images. We propose two novel detection methods: Our first method employs the magnitude spectrum of the input images to detect an adversarial attack. This simple and robust classifier can successfully detect adversarial perturbations of three commonly used attack methods. The second method builds upon the first and additionally extracts the phase of Fourier coefficients of feature-maps at different layers of the network. With this extension, we are able to improve adversarial detection rates compared to state-of-the-art detectors on five different attack methods.

The process of building and training Machine Learning models is always in the spotlight. There is a lot of talk about different Neural Network architectures, or new frameworks, facilitating the idea-to-implementation transition. While these are the heart of an ML engine, the circulatory system, which enables nutrients to move around and connects everything, is often missing. But what comprises this system? What gives the pipeline its pulse? The most important component in an ML pipeline works silently in the background and provides the glue that binds everything together.

The dream of autonomous vehicles is that they can avoid human error and save lives, but a new European Union Agency for Cybersecurity (ENISA) report has found that autonomous vehicles are "highly vulnerable to a wide range of attacks" that could be dangerous for passengers, pedestrians, and people in other vehicles. Attacks considered in the report include sensor attacks with beams of light, overwhelming object detection systems, back-end malicious activity, and adversarial machine learning attacks presented in training data or the physical world. "The attack might be used to make the AI'blind' for pedestrians by manipulating for instance the image recognition component in order to misclassify pedestrians. This could lead to havoc on the streets, as autonomous cars may hit pedestrians on the road or crosswalks," the report reads. "The absence of sufficient security knowledge and expertise among developers and system designers on AI cybersecurity is a major barrier that hampers the integration of security in the automotive sector."

If you follow VO discussion groups on social media, it's hard to miss the urgent and sometimes panicked tone among VOs when it comes to the subject of AI, or Artificial Intelligence voice overs. There seems to be a pervasive fear that we are all about to be replaced….sent Well, you may indeed go bald. There's nothing I can do about that. But, I can give you reassurance that the age of AI voiceover may in fact be to the benefit of serious professional voice actors.

The AI Council has published a "roadmap" of advice for government in respect of developing a UK state strategy for artificial intelligence (AI). Eye-catchingly, it advocates what it calls "moonshots" that "could tackle fundamental challenges such as creating'explainable AI' and developing smart materials for energy storage". The council is a non-statutory body chaired by Tabitha Goldstaub, consisting of 20 people from academia and industry, including Wendy Hall, professor of computer science at the University of Southampton, Marc Warner, the CEO of AI consultancy firm Faculty, and Adrian Smith, chief executive of The Alan Turing Institute. The council was launched in 2018, on the back of the government's response to a House of Lords AI report that recommended the UK pick ethics as a realistic niche in the related fields of artificial intelligence and machine learning. It was bolstered in 2019 with recruits from online retailer Ocado and the Independent Commission on Freedom of Information.