A common problem in Machine Learning and statistics consists in detecting whether the current sample in a stream of data belongs to the same distribution as previous ones, is an isolated outlier or inaugurates a new distribution of data. We present a hierarchical Bayesian algorithm that aims at learning a time-specific approximate posterior distribution of the parameters describing the distribution of the data observed. We derive the update equations of the variational parameters of the approximate posterior at each time step for models from the exponential family, and show that these updates find interesting correspondents in Reinforcement Learning (RL). In this perspective, our model can be seen as a hierarchical RL algorithm that learns a posterior distribution according to a certain stability confidence that is, in turn, learned according to its own stability confidence. Finally, we show some applications of our generic model, first in a RL context, next with an adaptive Bayesian Autoregressive model, and finally in the context of Stochastic Gradient Descent optimization.

Aiming at financial applications, we study the problem of learning the volatility under market microstructure noise. Specifically, we consider noisy discrete time observations from a stochastic differential equation and develop a novel computational method to learn the diffusion coefficient of the equation. We take a nonparametric Bayesian approach, where we model the volatility function a priori as piecewise constant. Its prior is specified via the inverse Gamma Markov chain. Sampling from the posterior is accomplished by incorporating the Forward Filtering Backward Simulation algorithm in the Gibbs sampler. Good performance of the method is demonstrated on two representative synthetic data examples. Finally, we apply the method on the EUR/USD exchange rate dataset.

Machine Learning techniques are widely used by online services (e.g. Google, Apple) in order to analyze and make predictions on user data. As many of the provided services are user-centric (e.g. personal photo collections, speech recognition, personal assistance), user data generated on personal devices is key to provide the service. In order to protect the data and the privacy of the user, federated learning techniques have been proposed where the data never leaves the user's device and "only" model updates are communicated back to the server. In our work, we propose a new threat model that is not concerned with learning about the content - but rather is concerned with the linkability of users during such decentralized learning scenarios. We show that model updates are characteristic for users and therefore lend themselves to linkability attacks. We show identification and matching of users across devices in closed and open world scenarios. In our experiments, we find our attacks to be highly effective, achieving 20x-175x chance-level performance. In order to mitigate the risks of linkability attacks, we study various strategies. As adding random noise does not offer convincing operation points, we propose strategies based on using calibrated domain-specific data; we find these strategies offers substantial protection against linkability threats with little effect to utility.

Reinforcement Learning (RL) can be extremely effective in solving complex, real-world problems. However, injecting human knowledge into an RL agent may require extensive effort and expertise on the human designer's part. To date, human factors are generally not considered in the development and evaluation of possible RL approaches. In this article, we set out to investigate how different methods for injecting human knowledge are applied, in practice, by human designers of varying levels of knowledge and skill. We perform the first empirical evaluation of several methods, including a newly proposed method named SASS which is based on the notion of similarities in the agent's state-action space. Through this human study, consisting of 51 human participants, we shed new light on the human factors that play a key role in RL. We find that the classical reward shaping technique seems to be the most natural method for most designers, both expert and non-expert, to speed up RL. However, we further find that our proposed method SASS can be effectively and efficiently combined with reward shaping, and provides a beneficial alternative to using only a single speedup method with minimal human designer effort overhead.

Recent research in AI is focusing towards generating narrative stories about visual scenes. It has the potential to achieve more human-like understanding than just basic description generation of images- in-sequence. In this work, we propose a solution for generating stories for images-in-sequence that is based on the Sequence to Sequence model. As a novelty, our encoder model is composed of two separate encoders, one that models the behaviour of the image sequence and other that models the sentence-story generated for the previous image in the sequence of images. By using the image sequence encoder we capture the temporal dependencies between the image sequence and the sentence-story and by using the previous sentence-story encoder we achieve a better story flow. Our solution generates long human-like stories that not only describe the visual context of the image sequence but also contains narrative and evaluative language. The obtained results were confirmed by manual human evaluation.

Although artificial intelligence (as a set of technologies, not in the sense of mimicking human intelligence) is here since a long time in many forms and ways, it's a term that quite some people, certainly IT vendors, don't like to use that much anymore – but artificial intelligence is real, for your business too. Instead of talking about artificial intelligence (AI) many describe the current wave of AI innovation and acceleration with – admittedly somewhat differently positioned – terms and concepts such as cognitive computing or focus on several real-life applications of artificial intelligence that often start with words such as "smart" (omni-present in anything related to the IoT as well), "intelligent", "predictive" and, indeed, "cognitive", depending on the exact application – and vendor. Despite the term issues, artificial intelligence is essential for and in, among others, information management, healthcare, life sciences, data analysis, digital transformation, security (cybersecurity and others), various consumer applications, next gen smart building technologies, FinTech, predictive maintenance, robotics and so much more. On top of that, AI is added to several other technologies, including IoT and big, as well as, small data analytics. There are many reasons why several vendors doubt using the term artificial intelligence for AI solutions/innovations and often package them in another term (trust us, we've been there). Artificial intelligence (AI) is a term that has somewhat of a negative connotation in general perception but also in the perception of technology leaders and firms.

SAN FRANCISCO – Alexa's been studying up on the royal wedding. So go ahead, ask the Amazon assistant everything your heart desires about the upcoming marriage of American actress Meghan Markle to Britain's Prince Harry. Amazon's artificial intelligence-infused Alexa digital assistant routinely gears up for blockbuster occasions, which the May 19 union of British and Hollywood royalty promises to be. Since November, the Amazon group that preps Alexa for news and cultural events has been adding answers and information so that no one gets the dreaded "Here's what I found on the web" answer in a response to questions ranging from "Who's going to walk Meghan Markle down the aisle?" to "What time is the royal wedding?" Google Assistant, with its Google Home rival to Alexa, isn't far behind when thrown the same types of questions. In our tests, Apple's Siri drew blanks on most of them, in terms of reading the answers aloud.

Subscribe to receive updates on Industry 4.0 The Fourth Industrial Revolution is upon us. The first three were based, respectively, on mechanization, mass production, and computing/automation; Industry 4.0 is all about the marriage of physical and digital technologies. Just as with the previous revolutions, Industry 4.0 is disrupting and redefining industries. This time, however, the revolution is progressing with unprecedented speed, driven by smart, connected technologies that are developing at an exponential rate.1 These technology innovations--including cloud computing and platform technologies, big data and analytics, mobile solutions, social and collaborative systems, Internet of Things (IoT) technology, and artificial intelligence (AI)--are fueling and accelerating a new era of digital business transformation. They're reshaping how organizations work, innovate, and create products--and enabling completely new kinds of products and services.2 They're spurring businesses to invent new business models and reimagine how they deliver value to their customers and markets. More broadly, industry boundaries are expanding and blurring, and relationships with business partners are being redefined. Yet too many organizations remain unprepared for the new revolution. A recent Deloitte Industry 4.0 study of C-level executives around the world indicates that, across all industries, only 14 percent of CXOs are "highly confident" that their organizations are ready to harness the changes associated with the new era.3

Elderly people in UK care homes will soon be looked after by robots that'learn and tailor their conversations' to who they are with. The £2.5 million ($3.4m) EU-funded trial scheme, starting in September, is designed help to take the strain off over-burdened carers, family and friends. The four-foot (1.2-metre) tall humanoid companions have been programmed to recognise the needs of residents, according to the firm behind the pilot. However, advocates for residents and their relatives have argued that it risks treating old people like commodities and losing the much needed'human touch'. The four-foot (1.2-metre) tall humanoid companions (pictured) are designed to recognise the needs of old people who might otherwise be alone The care-giving robots, called Pepper, have been designed by Japanese company Softbank Robotics.

The masters of cutting-edge AI technology recently granted me an advance visit to Amelia City, the not-yet-open lab and showplace for Amelia, IPsoft's notably humanesque AI. Amelia is so far best known for customer support/helpdesk applications (which is where my own clients, as a customer support and customer experience consultant, have encountered her to date) though her deployments have been expanding into other areas including medicine, HR, fundraising, and more. "Her" marquee clients already include some 20 of the Fortune 100, and the company is now developing pre-trained, limited-function mini-Amelias for small and medium-size businesses as well. Amelia City is at the very tip-end of Manhattan, with striking views of Ellis Island and the Statue of Liberty that are the mirror image of what immigrants starting a new life here would have seen. According to IPsoft CEO Chetan Dube (more from him soon), Amelia is going to bring nearly as dramatic a change to the inhabitants of our world: a new life both for better–economic results and freedom from drudgery–and worse–mass disruptions that will require workforce retraining and, probably, government interventions–and will do so with great inevitability. But let's not jump ahead.