This paper studies linear convergence of the subspace constrained mean shift (SCMS) algorithm, a well-known algorithm for identifying a density ridge defined by a kernel density estimator. By arguing that the SCMS algorithm is a special variant of a subspace constrained gradient ascent (SCGA) algorithm with an adaptive step size, we derive linear convergence of such SCGA algorithm. While the existing research focuses mainly on density ridges in the Euclidean space, we generalize density ridges and the SCMS algorithm to directional data. In particular, we establish the stability theorem of density ridges with directional data and prove the linear convergence of our proposed directional SCMS algorithm.

Across machine learning (ML) sub-disciplines, researchers make explicit mathematical assumptions in order to facilitate proof-writing. We note that, specifically in the area of fairness-accuracy trade-off optimization scholarship, similar attention is not paid to the normative assumptions that ground this approach. Such assumptions presume that 1) accuracy and fairness are in inherent opposition to one another, 2) strict notions of mathematical equality can adequately model fairness, 3) it is possible to measure the accuracy and fairness of decisions independent from historical context, and 4) collecting more data on marginalized individuals is a reasonable solution to mitigate the effects of the trade-off. We argue that such assumptions, which are often left implicit and unexamined, lead to inconsistent conclusions: While the intended goal of this work may be to improve the fairness of machine learning models, these unexamined, implicit assumptions can in fact result in emergent unfairness. We conclude by suggesting a concrete path forward toward a potential resolution.

While the predictive performance of modern statistical dependency parsers relies heavily on the availability of expensive expert-annotated treebank data, not all annotations contribute equally to the training of the parsers. In this paper, we attempt to reduce the number of labeled examples needed to train a strong dependency parser using batch active learning (AL). In particular, we investigate whether enforcing diversity in the sampled batches, using determinantal point processes (DPPs), can improve over their diversity-agnostic counterparts. Simulation experiments on an English newswire corpus show that selecting diverse batches with DPPs is superior to strong selection strategies that do not enforce batch diversity, especially during the initial stages of the learning process. Additionally, our diversityaware strategy is robust under a corpus duplication setting, where diversity-agnostic sampling strategies exhibit significant degradation.

Since its beginning in the 1950s, the field of artificial intelligence has cycled several times between periods of optimistic predictions and massive investment ("AI spring") and periods of disappointment, loss of confidence, and reduced funding ("AI winter"). Even with today's seemingly fast pace of AI breakthroughs, the development of long-promised technologies such as self-driving cars, housekeeping robots, and conversational companions has turned out to be much harder than many people expected. One reason for these repeating cycles is our limited understanding of the nature and complexity of intelligence itself. In this paper I describe four fallacies in common assumptions made by AI researchers, which can lead to overconfident predictions about the field. I conclude by discussing the open questions spurred by these fallacies, including the age-old challenge of imbuing machines with humanlike common sense.

With current AI technologies, harm done by AIs is limited to power that we put directly in their control. As said in [59], "For Narrow AIs, safety failures are at the same level of importance as in general cybersecurity, but for AGI it is fundamentally different." Despite AGI (artificial general intelligence) still being well out of reach, the nature of AI catastrophes has already changed in the past two decades. Automated systems are now not only malfunctioning in isolation, they are interacting with humans and with each other in real time. This shift has made traditional systems analysis more difficult, as AI has more complexity and autonomy than software has before. In response to this, we analyze how risks associated with complex control systems have been managed historically and the patterns in contemporary AI failures to what kinds of risks are created from the operation of any AI system. We present a framework for analyzing AI systems before they fail to understand how they change the risk landscape of the systems they are embedded in, based on conventional system analysis and open systems theory as well as AI safety principles. Finally, we present suggested measures that should be taken based on an AI system's properties. Several case studies from different domains are given as examples of how to use the framework and interpret its results.

Neural IR models have often been studied in homogeneous and narrow settings, which has considerably limited insights into their generalization capabilities. To address this, and to allow researchers to more broadly establish the effectiveness of their models, we introduce BEIR (Benchmarking IR), a heterogeneous benchmark for information retrieval. We leverage a careful selection of 17 datasets for evaluation spanning diverse retrieval tasks including open-domain datasets as well as narrow expert domains. We study the effectiveness of nine state-of-the-art retrieval models in a zero-shot evaluation setup on BEIR, finding that performing well consistently across all datasets is challenging. Our results show BM25 is a robust baseline and Reranking-based models overall achieve the best zero-shot performances, however, at high computational costs. In contrast, Dense-retrieval models are computationally more efficient but often underperform other approaches, highlighting the considerable room for improvement in their generalization capabilities. In this work, we extensively analyze different retrieval models and provide several suggestions that we believe may be useful for future work. BEIR datasets and code are available at https://github.com/UKPLab/beir.

Ryan Calo is the Lane Powell and D. Wayne Gittinger Professor at the University of Washington School of Law. He is a founding co-director of the interdisciplinary UW Tech Policy Lab and the UW Center for an Informed Public. Professor Calo holds adjunct appointments at the University of Washington Information School and the Paul G. Allen School of Computer Science and Engineering. The following is a lightly edited transcript of a discussion that took place shortly after the publication of the European Commission's proposed new regulation of artificial intelligence (AI). The European Commission has today released a proposed regulation around AI. This is obviously something that you have been prepared for and waiting to see happen. What did the EU put out? Years ago I wrote a primer and roadmap for AI policy and also hosted the inaugural Obama White House workshop on artificial intelligence policy. Many of the themes of that essay and of the workshop were reflected in the EU proposal, which is to say that they're not limiting themselves to decision-making by AI. Their approach is to look at the impacts of AI holistically, and to tackle everything from liability should there be harm, to additional obligations for high-risk uses, to facial recognition and biometrics.

Motorists could legally allow their cars to "self-drive" on British motorways later this year – but only slowly, the government has announced. Drivers could soon be allowed to read a newspaper or watch a film via the car's built-in screen in periods of slow-moving traffic, using automated lane-keeping system (Alks) technology that makes the car stay in lane and a safe distance from other vehicles. But insurers and motoring organisations said much more work needed to be done to ensure safety, after the Department for Transport confirmed it would pursue plans to allow new models fitted with Alks to drive without the driver's input. The cars will be defined as self-driving when the system is in operation, at a maximum speed of 37mph. According to the DfT, the technology, which will constantly monitor speed and distance from other cars, could improve road safety by reducing human error.

We have also seen that despite the discrepancies between countries from North to South, to Western countries to Eastern countries, there is a domino effect according to which major ethical issues and tendencies are almost simultaneously faced by every country at the same time, whatever their place in the AI race. It was the case for the AI tracking applications and the facial recognition applications during #COVID-19, and it will probably continue because AI is questioning the equilibrium of geopolitics worldwide. It also questions our ability to face fundamental and crucial questions as of the future of multilateralism. Below the translation of an article published in January, 2021 that highlights some trends I have foreseen in December 2020 for AI globally with some insights on the French market. Carrying out a prospective exercise is never easy and is even less so in the current context, which reminds us of the impermanence of all things and the need to adapt with agility, both individually and collectively, while keeping a long-term vision and without giving in to the call of falsely obvious and short-term choices.

Apple said it plans to spend $1 billion as it builds a new campus and engineering hub in the Research Triangle area of North Carolina, with most of the jobs expected to focus on machine learning, artificial intelligence, software engineering and other technology fields. It joins a $1 billion Austin, Texas campus announced in 2019. North Carolina's Economic Investment Committee on Monday approved a job-development grant that could provide Apple as much as $845.8 million in tax reimbursements over 39 years if Apple hits job and growth targets. State officials said the 3,000 jobs are expected to create $1.97 billion in new tax revenues to the state over the grant period. The iPhone maker said it would also establish a $100 million fund to support schools in the Raleigh-Durham area of North Carolina and throughout the state, as well as contribute $110 million to help build infrastructure such as broadband internet, roads, bridges and public schools in 80 North Carolina counties.