Last week I joined Toni Townes-Whitley, Josh Bersin and chief learning officers from around the world at the Microsoft Global Learning Summit to discuss the importance of future-proofing your workforce through learning and skill development. It was a pleasure to present to this group, and it reminded me of my own (very recent) experience in learning and skill development: When our CEO, Satya Nadella, asked me to lead HR at Microsoft. At that point, I'd led Microsoft Services for six years, and before that I held operations, strategy, management and development roles. What made Satya think I could take on a role as Chief People Officer for a global workforce of 115,000 people? We both took a leap of faith, and with his help, the help of the Microsoft Senior Leadership Team, my industry peers, and all 1,500 employees in HR at Microsoft, I have leaned into--and learned into--my role.

Visual analytics systems combine machine learning or other analytic techniques with interactive data visualization to promote sensemaking and analytical reasoning. It is through such techniques that people can make sense of large, complex data. While progress has been made, the tactful combination of machine learning and data visualization is still under-explored. This state-of-the-art report presents a summary of the progress that has been made by highlighting and synthesizing select research advances. Further, it presents opportunities and challenges to enhance the synergy between machine learning and visual analytics for impactful future research directions.

Asynchronous distributed machine learning solutions have proven very effective so far, but always assuming perfectly functioning workers. In practice, some of the workers can however exhibit Byzantine behavior, caused by hardware failures, software bugs, corrupt data, or even malicious attacks. We introduce \emph{Kardam}, the first distributed asynchronous stochastic gradient descent (SGD) algorithm that copes with Byzantine workers. Kardam consists of two complementary components: a filtering and a dampening component. The first is scalar-based and ensures resilience against $\frac{1}{3}$ Byzantine workers. Essentially, this filter leverages the Lipschitzness of cost functions and acts as a self-stabilizer against Byzantine workers that would attempt to corrupt the progress of SGD. The dampening component bounds the convergence rate by adjusting to stale information through a generic gradient weighting scheme. We prove that Kardam guarantees almost sure convergence in the presence of asynchrony and Byzantine behavior, and we derive its convergence rate. We evaluate Kardam on the CIFAR-100 and EMNIST datasets and measure its overhead with respect to non Byzantine-resilient solutions. We empirically show that Kardam does not introduce additional noise to the learning procedure but does induce a slowdown (the cost of Byzantine resilience) that we both theoretically and empirically show to be less than $f/n$, where $f$ is the number of Byzantine failures tolerated and $n$ the total number of workers. Interestingly, we also empirically observe that the dampening component is interesting in its own right for it enables to build an SGD algorithm that outperforms alternative staleness-aware asynchronous competitors in environments with honest workers.

We examine two fundamental tasks associated with graph representation learning: link prediction and semi-supervised node classification. We present a densely connected autoencoder architecture capable of learning a joint representation of both local graph structure and available external node features for the multi-task learning of link prediction and node classification. To the best of our knowledge, this is the first architecture that can be efficiently trained end-to-end in a single learning stage to simultaneously perform link prediction and node classification. We provide comprehensive empirical evaluation of our models on a range of challenging benchmark graph-structured datasets, and demonstrate significant improvement in accuracy over related methods for graph representation learning. Code implementation is available at https://github.com/vuptran/graph-representation-learning

Autonomous robots need to interact with unknown, unstructured and changing environments, constantly facing novel challenges. Therefore, continuous online adaptation for lifelong-learning and the need of sample-efficient mechanisms to adapt to changes in the environment, the constraints, the tasks, or the robot itself are crucial. In this work, we propose a novel framework for probabilistic online motion planning with online adaptation based on a bio-inspired stochastic recurrent neural network. By using learning signals which mimic the intrinsic motivation signal cognitive dissonance in addition with a mental replay strategy to intensify experiences, the stochastic recurrent network can learn from few physical interactions and adapts to novel environments in seconds. We evaluate our online planning and adaptation framework on an anthropomorphic KUKA LWR arm. The rapid online adaptation is shown by learning unknown workspace constraints sample-efficiently from few physical interactions while following given via points.

We are on the verge of a technological revolution that will fundamentally alter the way we live, work, and relate to one another unlike anything humankind has experienced before. The main driver for this technological revolution is Artificial Intelligence (AI). Technological change driven by AI will change not only what we do but also who we are. It will affect our identity and all the issues associated with it: our sense of privacy, our notions of ownership, our consumption patterns, the time we devote to work and leisure, and how we develop our careers, cultivate our skills, and nurture relationships. But the development and applications of artificial intelligence can also present a dystopian threat to our collective and individual well being. From SIRI to self-driving cars, artificial intelligence (AI) is progressing rapidly. While science fiction often portrays AI as robots with human-like characteristics, AI can encompass anything from Google's search algorithms to IBM's Watson to autonomous robots and weapons systems. Artificial intelligence today is often referred to as narrow AI (or weak AI), which is designed to perform a narrow task (eg:facial recognition or only internet searches or driving a car). The other kind of Artificial Intelligence is termed general AI (AGI or strong AI) which is designed to "think," and solve problems much like humans.

Stationed around the class, each robot has a mounted video screen controlled by the remote user that lets the student pan around the room to see and talk with the instructor and fellow students participating in-person. The study, published in Online Learning, found that robot learning generally benefits remote students more than traditional videoconferencing, in which multiple students are displayed on a single screen. Christine Greenhow, MSU associate professor of educational psychology and educational technology, said that instead of looking at a screen full of faces as she does with traditional videoconferencing, she can look a robot-learner in the eye -- at least digitally. "It was such a benefit to have people individually embodied in robot form -- I can look right at you and talk to you," Greenhow said. The technology, Greenhow added, also has implications for telecommuters working remotely and students with disabilities or who are ill.

AI teaching computers to make business sense of ill-lit 3D objects. We invested in Doxel, because of Saurabh Ladha, and his co-founder Robin Singh. They are without much parallel when it comes to the tech of 3D semantic understanding, and with their team of CS PhDs essentially writing software using AI to teach computers to make sense of the 3D world around them -- even when in less than ideal, real world sites that have little to no light. Both founders come with exceptionally strong engineering backgrounds, having met on the Dubai campus and then they split off to respectively Stanford and Ann Arbor Michigan for further education. This technology has broad application to industries of any kind wanting to know what's going on on any physical project of theirs, be it construction, agriculture, shipping, manufacture and many more have been relegated to a 2D static world.

One of the problems in demystifying artificial intelligence is that once an AI-based product reaches the public, "we stop calling it AI," said Tara Chklovski, the CEO and founder of Iridescent, a nonprofit that aims to educate and empower children and their parents on engineering and …

Their uses once seemed far off for companies, but with the introduction of Machine Learning-as-a-Service (MLaaS), data science is being brought to the masses. Machine learning, a branch of artificial intelligence, is the process of using self-iterating algorithms to analyse massive amounts of data by learning from the information and processing it with minimal supervision. Essentially, machines can learn from themselves through advanced algorithms data scientists create. This technology has implications across all fields, which is why financial institutions, health services, and more are all scrambling to hire skilled data scientists. Given the demand for machine learning services, MLaaS offerings have recently sprouted up to meet this need.