Graph models are relevant in many fields, such as distributed computing, intelligent tutoring systems or social network analysis. In many cases, such models need to take changes in the graph structure into account, i.e. a varying number of nodes or edges. Predicting such changes within graphs can be expected to yield important insight with respect to the underlying dynamics, e.g. with respect to user behaviour. However, predictive techniques in the past have almost exclusively focused on single edges or nodes. In this contribution, we attempt to predict the future state of a graph as a whole. We propose to phrase time series prediction as a regression problem and apply dissimilarity- or kernel-based regression techniques, such as 1-nearest neighbor, kernel regression and Gaussian process regression, which can be applied to graphs via graph kernels. The output of the regression is a point embedded in a pseudo-Euclidean space, which can be analyzed using subsequent dissimilarity- or kernel-based processing methods. We discuss strategies to speed up Gaussian Processes regression from cubic to linear time and evaluate our approach on two well-established theoretical models of graph evolution as well as two real data sets from the domain of intelligent tutoring systems. We find that simple regression methods, such as kernel regression, are sufficient to capture the dynamics in the theoretical models, but that Gaussian process regression significantly improves the prediction error for real-world data.

A Google engineer who got fired over a controversial memo that criticized the company's diversity policies said that there might be biological reasons there are fewer women engineers. But top computer science schools have proven that a few cultural changes can increase the number of women in the field. In 2006, only about 10 percent of computer science majors at Harvey Mudd College were women. That's pretty low since Harvey Mudd is a school for students who are interested in science, math and technology. Then, Maria Klawe began her tenure as president of the college.

Mapping 1,100 years of metaphoric English language, researchers at UC Berkeley and Lehigh University in Pennsylvania have detected patterns in how English speakers have added figurative word meanings to their vocabulary. Researchers called the original semantic domain the "source domain" and the domain that the metaphorical meaning was extended to, the "target domain." More than 1,400 online participants were recruited to rate semantic domains such as "water" or "mind" according to the degree to which they were related to the external world (light, plants), animate things (humans, animals), or intense emotions (excitement, fear). In comparing their computational predictions against the actual historical record provided by the Metaphor Map of English, researchers found that their models correctly forecast about 75 percent of recorded metaphorical language mappings over the past millennium.

Mapping 1,100 years of metaphoric English language, researchers at UC Berkeley and Lehigh University in Pennsylvania have detected patterns in how English speakers have added figurative word meanings to their vocabulary. Researchers called the original semantic domain the "source domain" and the domain that the metaphorical meaning was extended to, the "target domain." More than 1,400 online participants were recruited to rate semantic domains such as "water" or "mind" according to the degree to which they were related to the external world (light, plants), animate things (humans, animals), or intense emotions (excitement, fear). In comparing their computational predictions against the actual historical record provided by the Metaphor Map of English, researchers found that their models correctly forecast about 75 percent of recorded metaphorical language mappings over the past millennium.

We arrive a few minutes late, thanks to a traffic jam that is now routine on the Outer Ring Road in Bengaluru. In the lobby of a hotel in Cessna Business Park, which houses Cisco, Flipkart and InMobi, D J Patil settles into a chat with my colleague Sriram Sharma. Patil comes across as an easygoing guy for a scientist. He was the first chief data scientist at the White House, handpicked by then US president Barack Obama. Patil and Jeff Hammerbacher coined the term "data scientist" in 2008.

Alex Lacamoire has hearing loss. But the Tony-winning music director of "Hamilton" wants you to know, he's no Beethoven. He's heard that you can see teeth marks on the wood inside Beethoven's piano "because he would bite it to try to be able to hear the vibrations," Lacamoire said. My hearing is not that bad." When he was 2, growing up near Los Angeles' Koreatown, Lacamoire would sit in front of the stereo and stare into the speaker, drawn to music like a drug. When he was 3, his mother observed him sitting too close to the TV, following the characters on "Sesame Street" with his eyes. "I noticed that when I called him, he'd run away like he wasn't paying attention," Maria Lacamoire said. She took him for a hearing test, where it was discovered that he had mild hearing loss. "I think I was a little bit too young for it to really understand," Lacamoire said. "All I remember is, like, oh wow, they're putting this weird goop in my ear to mold me [for hearing aids] and then I walked away and I had these little apparatuses behind my ears." When he was 6, the school district recommended that Lacamoire attend a special class that combined sign-language instruction along with spoken language. "That was devastating for me," his mother said, "because I didn't notice any other problem with him, because he was very smart." She appealed the decision, and Lacamoire was given an IQ test. He not only joined a mainstream class at Commonwealth Avenue Elementary School but also skipped the first grade. "Alex was the most outstanding student I ever had," said his second-grade teacher, Dorothy Chapman, who taught at Commonwealth for 25 years and retired in 2002. Children with hearing loss, especially when that loss is identified late, often lag behind their peers because they've absorbed less vocabulary and less information. Chapman said the charming little 6-year-old would finish his assignments in five minutes, whereas it took his classmates 20, so she would give him third-grade work. "I've just always been drawn to design, whether it's uniformity or harmony -- and by harmony I mean symmetry and balance and those kinds of things," Lacamoire said. He found beauty and design in the piano, and starting lessons at age 4. After his family moved to Miami when he was 9, he attended an arts high school and then the New World School of the Arts. For Lacamoire, music was "as fluid to me as writing down words.

About this course: If you want to break into cutting-edge AI, this course will help you do so. Deep learning engineers are highly sought after, and mastering deep learning will give you numerous new career opportunities. Deep learning is also a new "superpower" that will let you build AI systems that just weren't possible a few years ago. In this course, you will learn the foundations of deep learning. When you finish this class, you will: - Understand the major technology trends driving Deep Learning - Be able to build, train and apply fully connected deep neural networks - Know how to implement efficient (vectorized) neural networks - Understand the key parameters in a neural network's architecture This course also teaches you how Deep Learning actually works, rather than presenting only a cursory or surface-level description.

The current processes for building machine learning systems require practitioners with deep knowledge of machine learning. This significantly limits the number of machine learning systems that can be created and has led to a mismatch between the demand for machine learning systems and the ability for organizations to build them. We believe that in order to meet this growing demand for machine learning systems we must significantly increase the number of individuals that can teach machines. We postulate that we can achieve this goal by making the process of teaching machines easy, fast and above all, universally accessible. While machine learning focuses on creating new algorithms and improving the accuracy of "learners", the machine teaching discipline focuses on the efficacy of the "teachers". Machine teaching as a discipline is a paradigm shift that follows and extends principles of software engineering and programming languages. We put a strong emphasis on the teacher and the teacher's interaction with data, as well as crucial components such as techniques and design principles of interaction and visualization. In this paper, we present our position regarding the discipline of machine teaching and articulate fundamental machine teaching principles. We also describe how, by decoupling knowledge about machine learning algorithms from the process of teaching, we can accelerate innovation and empower millions of new uses for machine learning models.

Scale of data and scale of computation infrastructures together enable the current deep learning renaissance. However, training large-scale deep architectures demands both algorithmic improvement and careful system configuration. In this paper, we focus on employing the system approach to speed up large-scale training. Via lessons learned from our routine benchmarking effort, we first identify bottlenecks and overheads that hinter data parallelism. We then devise guidelines that help practitioners to configure an effective system and fine-tune parameters to achieve desired speedup. Specifically, we develop a procedure for setting minibatch size and choosing computation algorithms. We also derive lemmas for determining the quantity of key components such as the number of GPUs and parameter servers. Experiments and examples show that these guidelines help effectively speed up large-scale deep learning training.

More than half a century later, the disciplines have graduated from the theoretical to practical, real world applications. We'll have some of the top minds in both categories to discuss the latest advances and future of AI and ML on stage and Disrupt San Francisco in just over a month. We'll be joined on stage by Brian Krzanich of Intel, John Giannandrea of Google, Sebastian Thrun of Udacity and Andrew Ng of Baidu, to outline the various ways these cutting edge technologies are already impacting our lives, from simple smart assistants, to self-driving cars. It's a broad range of speakers, which is good news, because we've got a lot of ground to cover in some of the industry's most exciting advances. John (JG) Giannandrea, SVP Engineering at Google: Giannandrea joined Google in 2010, when the company acquired his startup Metaweb Technologies, a move that formed the basis for the search giant's Knowledge Graph technology.