Neural Machine Translation (NMT) generates target words sequentially in the way of predicting the next word conditioned on the context words. At training time, it predicts with the ground truth words as context while at inference it has to generate the entire sequence from scratch. This discrepancy of the fed context leads to error accumulation among the way. Furthermore, word-level training requires strict matching between the generated sequence and the ground truth sequence which leads to overcorrection over different but reasonable translations. In this paper, we address these issues by sampling context words not only from the ground truth sequence but also from the predicted sequence by the model during training, where the predicted sequence is selected with a sentence-level optimum. Experiment results on Chinese->English and WMT'14 English->German translation tasks demonstrate that our approach can achieve significant improvements on multiple datasets.

We establish a theoretical link between adversarial training and operator norm regularization for deep neural networks. Specifically, we show that adversarial training is a data-dependent generalization of spectral norm regularization. This intriguing connection provides fundamental insights into the origin of adversarial vulnerability and hints at novel ways to robustify and defend against adversarial attacks. We provide extensive empirical evidence to support our theoretical results.

Most state-of-the-art machine learning (ML) classification systems are vulnerable to adversarial perturbations. As a consequence, adversarial robustness poses a significant challenge for the deployment of ML-based systems in safety- and security-critical environments like autonomous driving, disease detection or unmanned aerial vehicles. In the past years we have seen an impressive amount of publications presenting more and more new adversarial attacks. However, the attack research seems to be rather unstructured and new attacks often appear to be random selections from the unlimited set of possible adversarial attacks. With this publication, we present a structured analysis of the adversarial attack creation process. By detecting different building blocks of adversarial attacks, we outline the road to new sets of adversarial attacks. We call this the "attack generator". In the pursuit of this objective, we summarize and extend existing adversarial perturbation taxonomies. The resulting taxonomy is then linked to the application context of computer vision systems for autonomous vehicles, i.e. semantic segmentation and object detection. Finally, in order to prove the usefulness of the attack generator, we investigate existing semantic segmentation attacks with respect to the detected defining components of adversarial attacks.

We consider the black-box adversarial setting, where the adversary has to generate adversarial perturbations without access to the target models to compute gradients. Previous methods tried to approximate the gradient either by using a transfer gradient of a surrogate white-box model, or based on the query feedback. However, these methods often suffer from low attack success rates or poor query efficiency since it is non-trivial to estimate the gradient in a high-dimensional space with limited information. To address these problems, we propose a prior-guided random gradient-free (P-RGF) method to improve black-box adversarial attacks, which takes the advantage of a transfer-based prior and the query information simultaneously. The transfer-based prior given by the gradient of a surrogate model is appropriately integrated into our algorithm by an optimal coefficient derived by a theoretical analysis. Extensive experiments demonstrate that our method requires much fewer queries to attack black-box models with higher success rates compared with the alternative state-of-the-art methods.

San Francisco supervisors approved a ban on police using facial recognition technology, making it the first city in the U.S. with such a restriction. SAN FRANCISCO – A routine traffic stop goes dangerously awry when a police officer's body camera uses its built-in facial recognition software to misidentify a motorist as a convicted felon. At best, lawsuits are launched. That imaginary scenario is what some California lawmakers are trying to avoid by supporting Assembly Bill 1215, the Body Camera Accountability Act, which would ban the use of facial recognition software in police body cams – a national first if it passes a Senate vote this summer and is signed by Gov. Gavin Newsom. State law enforcement officials here do not now employ the technology to scan those in the line of sight of officers.

HONG KONG (Reuters Breakingviews) - Artificial intelligence doesn't hate you, prominent researcher Eliezer Yudkowsky wrote, "nor does it love you, but you are made of atoms which it can use for something else". This sets the scene for Tom Chivers' fascinating new book, which borrows its title from the quote, on why so-called superintelligence should be viewed as an existential threat potentially greater than nuclear weapons or climate change. The "strange, irascible and brilliant" Yudkowsky is a central figure throughout the book. His early musings on the potential and dangers of artificial intelligence during the mid- to late-2000s gave birth to the Rationalist movement, a loose community dedicated to AI safety. Chivers, a former science journalist with Buzzfeed and the Telegraph, offers a meticulously researched investigation into who the Rationalists are, and more importantly why they believe humanity is fast approaching an inflection point between "extinction and godhood".

The use of multilingual translation tools is expanding in Japan, where foreign workers are expected to increase in the wake of April's launch of new visa categories. A growing number of local governments, labor unions and other entities have decided to introduce translation tools, which can help foreigners when going through administrative procedures as they allow local officials and other officers to talk to such applicants in their mother languages. "Talking in the applicants' own languages makes it easier to convey our cooperative stance," said an official in Tokyo's Sumida Ward. The ward introduced VoiceBiz, an audio translation app developed by Toppan Printing Co. that covers 30 languages. The app, which can be downloaded onto smartphones and tablet computers, will be used in eight municipalities, including Osaka and Ayase in Kanagawa Prefecture, company officials said.

A survey with 83 Gartner Research Circle members indicates that, among 35% of the respondents, "identifying use cases for AI" was the top three challenges in exploring and adopting AI. It's impossible to recommend a single use case that is applicable for every city, because different cities have different priorities for their smart city projects. Among all the IoT use cases in smart cities, which keep evolving and expanding, ensure you give priority to those use cases of higher value. How can the value of use cases be defined in a smart city context then? There are some general principles to follow based on two key parameters: value that the project would bring to the citizens and value that the project would deliver for the governments.

Most of us do not have an equal voice or representation in this new world order. Leading the way instead are scientists and engineers who don't seem to understand how to represent how we live as individuals or in groups--the main ways we live, work, cooperate, and exist together--nor how to incorporate into their models our ethnic, cultural, gender, age, geographic or economic diversity, either. The result is that AI will benefit some of us far more than others, depending upon who we are, our gender and ethnic identities, how much income or power we have, where we are in the world, and what we want to do. The power structures that developed the world's complex civic and corporate systems were not initially concerned with diversity or equality, and as these systems migrate to becoming automated, untangling and teasing out the meaning for the rest of us becomes much more complicated. In the process, there is a risk that we will become further dependent on systems that don't represent us.

AI Ethics is now a global topic of discussion in academic and policy circles. At least 63 public-private initiatives have produced statements describing high-level principles, values, and other tenets to guide the ethical development, deployment, and governance of AI. According to recent meta-analyses, AI Ethics has seemingly converged on a set of principles that closely resemble the four classic principles of medical ethics. Despite the initial credibility granted to a principled approach to AI Ethics by the connection to principles in medical ethics, there are reasons to be concerned about its future impact on AI development and governance. Significant differences exist between medicine and AI development that suggest a principled approach in the latter may not enjoy success comparable to the former. Compared to medicine, AI development lacks (1) common aims and fiduciary duties, (2) professional history and norms, (3) proven methods to translate principles into practice, and (4) robust legal and professional accountability mechanisms. These differences suggest we should not yet celebrate consensus around high-level principles that hide deep political and normative disagreement.