We propose a framework, named Aggregated Wasserstein, for computing a dissimilarity measure or distance between two Hidden Markov Models with state conditional distributions being Gaussian. For such HMMs, the marginal distribution at any time position follows a Gaussian mixture distribution, a fact exploited to softly match, aka register, the states in two HMMs. We refer to such HMMs as Gaussian mixture model-HMM (GMM-HMM). The registration of states is inspired by the intrinsic relationship of optimal transport and the Wasserstein metric between distributions. Specifically, the components of the marginal GMMs are matched by solving an optimal transport problem where the cost between components is the Wasserstein metric for Gaussian distributions. The solution of the optimization problem is a fast approximation to the Wasserstein metric between two GMMs. The new Aggregated Wasserstein distance is a semi-metric and can be computed without generating Monte Carlo samples. It is invariant to relabeling or permutation of states. The distance is defined meaningfully even for two HMMs that are estimated from data of different dimensionality, a situation that can arise due to missing variables. This distance quantifies the dissimilarity of GMM-HMMs by measuring both the difference between the two marginal GMMs and that between the two transition matrices. Our new distance is tested on tasks of retrieval, classification, and t-SNE visualization of time series. Experiments on both synthetic and real data have demonstrated its advantages in terms of accuracy as well as efficiency in comparison with existing distances based on the Kullback-Leibler divergence.

Effective techniques for analyzing and detecting changes in streaming data, especially in the era of big data, pose new challenges to the machine learning and the statistics community [1], [2]. As a result, early approaches for detecting statistical changes in a time series (such as change point detection), have had to be extended for online detection of changes in a multivariate data streams [3], [4]. Some of these techniques for detecting the intrinsic change in the relationship of the incoming data streams have been applied to numerous real-world applications, such as fraud detection, user preference prediction and email filtering, [5], [6]. Online classification is another common task performed on streaming multivariate time series data that takes advantage of these statistical relationships to predict a class label at each time index [7]. If the underlying source generating the data is not stationary, the optimal decision rule for the classifier would change over time - a phenomena known as concept drift [8]. Given the impact of concept drift on the predictive performance of an online classifier, there is a need to detect these concept drifts as early as possible. The inability of change point detection approaches to detect these concept drifts, has motivated the need for concept drift detection approaches that not only monitor the join distribution of a multivariate data stream but also changes in its relationship to the class labels of the streaming data. Shujian Yu and José C. Príncipe are with the Department of Electrical and Computer Engineering, University of Florida, Gainesville, FL 32611, USA.

Thank you all for the huge response to this emerging course! We are delighted to have over 300 students in over 145 different countries. I'm genuinely touched by the overwhelmingly positive and thoughtful reviews. It's such a privilege to share and introduce this important topic with everyday people in a clear and understandable way. I'm also excited to announce that I have created real closed captions for all course material, so weather you need them due to a hearing impairment, or find it easier to follow long (great for ESL students!)... I've got you covered.

Amazon yesterday announced its ONNX-MXNet package to import Open Neural Network Exchange (ONNX) deep learning models into Apache MXNet, signifying the company is on-board with Facebook and Microsoft in efforts to open-source AI. With the ONNX-MXNet Python package, developers running models based on open-source ONNX will be able to run them on Apache MXNet. Basically, this allows AI developers to keep models but switch networks, as opposed to starting from scratch. If you can imagine a thousand start ups and another thousand universities all creating at the bleeding edge of machine learning technology, but unable to share work due to'format' issues, you won't be very far off from the state of things without initiatives like ONNX. With Facebook and Microsoft all-in on the idea of open-source AI platforms, and now Amazon joining them, it's looking like ONNX is the path forward.

Our world is advancing at an extremely rapid rate. Technologies such as artificial intelligence, machine learning, drones, internet of things, augmented reality, and blockchain are growing in popularity every single day. Personally, I feel another industrial revolution is approaching quickly and the world we will in is going to drastically change. Blockchain is a difficult technology to understand but it has the potential to impact many organizations across the globe. If you're looking to get a head start on an innovative idea that will change our world then you're in the right place!

We aim to enable robot to learn object manipulation by imitation. Given external observations of demonstrations on object manipulations, we believe that two underlying problems to address in learning by imitation is 1) segment a given demonstration into skills that can be individually learned and reused, and 2) formulate the correct RL (Reinforcement Learning) problem that only considers the relevant aspects of each skill so that the policy for each skill can be effectively learned. Previous works made certain progress in this direction, but none has taken private information into account. The public information is the information that is available in the external observations of demonstration, and the private information is the information that are only available to the agent that executes the actions, such as tactile sensations. Our contribution is that we provide a method for the robot to automatically segment the demonstration of object manipulations into multiple skills, and formulate the correct RL problem for each skill, and automatically decide whether the private information is an important aspect of each skill based on interaction with the world. Our experiment shows that our robot learns to pick up a block, and stack it onto another block by imitating an observed demonstration. The evaluation is based on 1) whether the demonstration is reasonably segmented, 2) whether the correct RL problems are formulated, 3) and whether a good policy is learned.

If you want a talk on Python with the Theano library in under an hour, targeted towards beginners, then you can refer to this talk by Alec Radford. Unlike most other talks on this topic, this one compares the features of an'old' net versus a'modern' net, ie nets prior to 2000 versus nets post-2012.

Artificial intelligence (AI) is rapidly transforming industries, enabling new experiences, greater efficiencies and cost savings. As advances in AI continue, few areas of our lives will remain untouched, education included. With the explosion of new technologies, education is becoming more affordable and accessible and innovators are pushing the boundaries of what is possible. The student experience is changing, as high levels of customisation are made possible. However, this technology has a limit, beyond which the human element is an essential part of the learning experience.

Video Friday is your weekly selection of awesome robotics videos, collected by your Automaton bloggers. We'll also be posting a weekly calendar of upcoming robotics events for the next two months; here's what we have so far (send us your events!): Let us know if you have suggestions for next week, and enjoy today's videos. With a title like "What's new, Atlas?" for a video like this, you know that Boston Dynamics is just messing with us now: The game played out on a real set wherein Cozmo would roll through a series of trials, exploring rooms and solving puzzles which tested his ability to move, to place, stack, and turn blocks, and recognize faces and pets, testing Reddit's collective will to help him. Cozmo's quest: to gather three golden key cubes to be able to escape to Reddit's front page.

For as long as artificial intelligence and machine learning tools have been moving into the workforce, there have been rumblings of robots taking over the work of people, and the impact that could have on their career prospects. However, new studies undertaken by global professional services brand Genpact of 5,000 respondents in the United Kingdom, United States, and Australia, shows that the level of concern among the workers themselves is not very high. Roughly twenty percent of those surveyed in the UK felt that their jobs were threatened by AI, with only six percent feeling this strongly. But, although they did not feel overly cautious about their own prospects, they saw the potential disadvantages for the next generation of workers, with over fifty percent responding there was a threat to their children's careers, and over eighty percent stating that new skills will be needed for those workers in order to succeed in an AI advanced environment. The reason for this caution can be found in the training, or lack thereof, in the use of AI.