According to the latest research from Counterpoint's Components Tracker Service, one in three smartphones to be shipped in 2020 will natively embed machine learning and artificial intelligence (AI) capabilities at the chipset level. Apple, with its Bionic system on chip (SoC), proliferating across its complete portfolio over the next couple of years, will drive native AI adoption in smartphones. Its universal adoption of AI-capable SoCs will likely enable Apple to lead the AI-capable chip market through 2020. Huawei, with its HiSilicon Kirin 970 SoC, launched in September and finding application in the Huawei Mate 10 series launched today in Munich, is second to market after Apple with AI-capable smartphones. The Huawei Mate 10 is able to accomplish diverse computational tasks efficiently, thanks to the neural processing unit at the heart of the Kirin 970 SoC.

Accumulating evidence suggest that human behavior in trial-and-error learning tasks based on decisions between discrete actions may involve a combination of reinforcement learning (RL) and working-memory (WM). While the understanding of brain activity at stake in this type of tasks often involve the comparison with non-human primate neurophysiological results, it is not clear whether monkeys use similar combined RL and WM processes to solve these tasks. Here we analyzed the behavior of five monkeys with computational models combining RL and WM. Our model-based analysis approach enables to not only fit trial-by-trial choices but also transient slowdowns in reaction times, indicative of WM use. We found that the behavior of the five monkeys was better explained in terms of a combination of RL and WM despite inter-individual differences. The same coordination dynamics we used in a previous study in humans best explained the behavior of some monkeys while the behavior of others showed the opposite pattern, revealing a possible different dynamics of WM process. We further analyzed different variants of the tested models to open a discussion on how the long pretraining in these tasks may have favored particular coordination dynamics between RL and WM. This points towards either inter-species differences or protocol differences which could be further tested in humans.

Traditional dictionary learning methods are based on quadratic convex loss function and thus are sensitive to outliers. In this paper, we propose a generic framework for robust dictionary learning based on concave losses. We provide results on composition of concave functions, notably regarding super-gradient computations, that are key for developing generic dictionary learning algorithms applicable to smooth and non-smooth losses. In order to improve identification of outliers, we introduce an initialization heuristic based on undercomplete dictionary learning. Experimental results using synthetic and real data demonstrate that our method is able to better detect outliers, is capable of generating better dictionaries, outperforming state-of-the-art methods such as K-SVD and LC-KSVD.

We study a variant of the stochastic multi-armed bandit (MAB) problem in which the rewards are corrupted. In this framework, motivated by privacy preservation in online recommender systems, the goal is to maximize the sum of the (unobserved) rewards, based on the observation of transformation of these rewards through a stochastic corruption process with known parameters. We provide a lower bound on the expected regret of any bandit algorithm in this corrupted setting. We devise a frequentist algorithm, KLUCB-CF, and a Bayesian algorithm, TS-CF and give upper bounds on their regret. We also provide the appropriate corruption parameters to guarantee a desired level of local privacy and analyze how this impacts the regret. Finally, we present some experimental results that confirm our analysis.

Computational models in fields such as computational neuroscience are often evaluated via stochastic simulation or numerical approximation. Fitting these models implies a difficult optimization problem over complex, possibly noisy parameter landscapes. Bayesian optimization (BO) has been successfully applied to solving expensive black-box problems in engineering and machine learning. Here we explore whether BO can be applied as a general tool for model fitting. First, we present a novel hybrid BO algorithm, Bayesian adaptive direct search (BADS), that achieves competitive performance with an affordable computational overhead for the running time of typical models. We then perform an extensive benchmark of BADS vs. many common and state-of-the-art nonconvex, derivative-free optimizers, on a set of model-fitting problems with real data and models from six studies in behavioral, cognitive, and computational neuroscience. With default settings, BADS consistently finds comparable or better solutions than other methods, including `vanilla' BO, showing great promise for advanced BO techniques, and BADS in particular, as a general model-fitting tool.

The Multi-Armed Bandits (MAB) framework highlights the tradeoff between acquiring new knowledge (Exploration) and leveraging available knowledge (Exploitation). In the classical MAB problem, a decision maker must choose an arm at each time step, upon which she receives a reward. The decision maker's objective is to maximize her cumulative expected reward over the time horizon. The MAB problem has been studied extensively, specifically under the assumption of the arms' rewards distributions being stationary, or quasi-stationary, over time. We consider a variant of the MAB framework, which we termed Rotting Bandits, where each arm's expected reward decays as a function of the number of times it has been pulled. We are motivated by many real-world scenarios such as online advertising, content recommendation, crowdsourcing, and more. We present algorithms, accompanied by simulations, and derive theoretical guarantees.

Google announced today that its virtual assistant will now support two new languages -- Spanish for users in the US, Mexico and Spain as well as Italian for those in Italy. These languages will be rolling out over the next few weeks, according to Google, and while initially they'll only be available for Android, the company says iPhone support will be released later this year. With these additions, Google Assistant now supports a total of eight languages -- including English, Portuguese, French, German, Japanese and Korean -- in 12 countries around the world. While that puts Google's virtual assistant in line with Microsoft's Cortana, which also supports eight languages, ahead of Amazon's Alexa, which only supports German and a few versions of English, and ahead of Samsung's Bixby, which supports just English and Korean, it's still well behind Apple's Siri. Apple's assistant currently supports 20 languages -- quite a few more than any of its counterparts.

The 49th Sveriges Riksbank prize in economic sciences – commonly referred to as the Nobel Prize for economics – has been awarded to Richard H Thaler for his contributions to behavioural economics. He was a key proponent of the idea that humans do not act entirely rationally and is primarily known for his often misunderstood concept of Nudge Theory. Attributed to Richard Thaler and Cass Sunstein in their book "Nudge: Improving Decisions about Health, Wealth, and Happiness," a "Nudge" can change the behaviour or decision that a human will make. The basis of Nudge theory is to apply an understanding of predicted behaviours to shape and influence that automated process. Leveraging a variety of different strategies, such as default settings, information as an incentive and right context, companies have proven the ability to change someone's behaviour through a successful Nudge.

Microsoft hopes to take a dedicated AI chip that it's building into the next generation of its HoloLens augmented-reality headset and push it into other hardware, presumably PCs and devices built by other countries. Speaking to CNBC, Microsoft's device chief Panos Panay said the company continues to work on an AI chipset for the next generation of the HoloLens, which so far has been sold primarily to commercial partners. According to Panay, Microsoft is poised to expand sales of the HoloLens from 10 markets to a total of 39, adding 29 new markets across Western Europe. Microsoft disclosed that it was building the new AI chipset in July, though it was unclear what functions the chipset would handle. Microsoft said then that the more traditional approaches used in the original HoloLens and other devices enhanced existing cloud computing fabrics, but that it was seeking to develop a more independent chip that could be used either online or offline.

When you sign up to labor in the "Mental Work" factory, you're equipped with a brain-scanning headset and taught how to use it. The headset uses standard EEG electrodes to record your brainwaves, and the associated software can pick out specific patterns. The factory overseer explains that this brain-computer interface has been programmed to respond to a neural pattern that occurs when you imagine squeezing a ball in your hand. Then you're introduced to the machines you'll be controlling. They are things of beauty, made of lightweight aluminum and finished in chrome.