We propose and analyze a regularization approach for structured prediction problems. We characterize a large class of loss functions that allows to naturally embed structured outputs in a linear space. We exploit this fact to design learning algorithms using a surrogate loss approach and regularization techniques. We prove universal consistency and finite sample bounds characterizing the generalization properties of the proposed methods. Experimental results are provided to demonstrate the practical usefulness of the proposed approach.

This paper proposes an out-of-sample extension framework for a global manifold learning algorithm (Isomap) that uses temporal information in out-of-sample points in order to make the embedding more robust to noise and artifacts. Given a set of noise-free training data and its embedding, the proposed framework extends the embedding for a noisy time series. This is achieved by adding a spatio-temporal compactness term to the optimization objective of the embedding. To the best of our knowledge, this is the first method for out-of-sample extension of manifold embeddings that leverages timing information available for the extension set. Experimental results demonstrate that our out-of-sample extension algorithm renders a more robust and accurate embedding of sequentially ordered image data in the presence of various noise and artifacts when compared to other timing-aware embeddings. Additionally, we show that an out-of-sample extension framework based on the proposed algorithm outperforms the state of the art in eye-gaze estimation.

Industrial IoT startup, Petasense was selected by leading venture capital firms, Data Collective and Emergence Capital, as one of the ten most disruptive machine learning startups out of over 350 contenders. The process was highly competitive with a selection rate of under 3%. This puts Petasense amongst the most promising startups implementing machine learning to solve an important real world problem. Petasense, along with nine others, was invited to participate in Google's machine learning event at the Google Launchpad in San Francisco on Wednesday, July 12, 2017. All ten companies were awarded $200,000 in Google Cloud Platform credits and an opportunity to pitch to an exclusive group of VCs, and experts in the field of machine learning and artificial intelligence.

Intel Corp. more than doubled its second-quarter profit as sales of its personal computer chips strengthened and the company made further inroads in promising new areas of technology. The world's largest chipmaker also brightened its outlook for the remainder of the year. The report released Thursday drew a lukewarm reaction from investors as Intel's stock edged up 13 cents to $35.10 in extended trading. Intel earned $2.81 billion, or 58 cents per share during the three-month period ended July 1. That compared to net income of $1.33 billion, or 27 cents per share, at the same time last year.

Course Description Caffe is a deep learning framework made with expression, speed, and modularity in mind. Audience This course is suitable for Deep Learning researchers and engineers interested in utilizing Caffe as a framework. After completing this course, delegates will be able to: understand Caffe's structure and deployment mechanisms carry out installation / production environment / architecture tasks and configuration assess code quality, perform debugging, monitoring implement advanced production like training models, implementing layers and logging

In the U.S., 3.6 out of 1000 school-aged children are diagnosed with cerebral palsy (CP). Their symptoms include abnormal gait patterns which results in joint degeneration over time. Slow walking speed, reduced range of motion of the joints, small step length, large body sway, and absence of a heel strike are other difficulties that children with CP experience. A subset of these children exhibit crouch gait which is characterized by excessive flexion of the hips, knees, or ankles. A team led by Sunil Agrawal, professor of mechanical engineering and of rehabilitation and regenerative medicine at Columbia Engineering, has published a pilot study in Science Robotics that demonstrates a robotic training method that improves posture and walking in children with crouch gait by enhancing their muscle strength and coordination.

The Impact of AI and Machine Learning on our Jobs Over the past year there have been an increasing number of articles written about jobs that can be done by a machine versus a person. I tend to be pretty optimistic about the future, but I don't believe anyone can know how the nature of jobs will be transformed as automation is introduced into various aspects of life. Here's an article from Fast Company that appeared 3 1/2 years ago about the changes coming from machine learning and artificial intelligence technologies. I don't think it's aged well. Here are three things it listed: 1) Unstructured problem-solving: solving for problems in which the rules do not currently exist.

IMAGE: This is the overview of Pharma AI drug discovery pipeline. Thursday, July 27, 2017, Baltimore, Md., Insilico Medicine ("Insilico"), a Baltimore-based leader in artificial intelligence ("AI") for drug discovery and biomarker development, is pleased to announce a multi-year drug development agreement with the biotechnology company Juvenescence AI Limited ("Juvenescence AI"). Juvenescence AI will develop the first compounds generated by Insilico's deep-learned drug discovery engines, which train over structural, functional, and phenotypic data in order to predict the biological activity of compounds. Insilico's platforms incorporate new AI techniques such as Generative Adversarial Networks in order to generate novel compounds with desired pharmacokinetic and pharmacodynamic properties. As part of the agreement, Juvenescence Limited ("Juvenescence"), the parent company of Juvenescence AI, made a direct investment into Insilico to further advance Insilico's drug discovery platform and develop a set of companion multi-modal disease biomarkers.

South Korean professional Go player Lee Sedol is seen on the screen during the Google DeepMind Challenge Match against Google's artificial intelligence program, AlphaGo, in Seoul, South Korea, March 9, 2016. Lee Sedol lost the first match. LOS ANGELES, July 27 (Xinhua) -- The war between tech titans has begun. Some are worried about what artificial intelligence (AI) will mean for humanity, calling for slowing down the process of building it, while others are pretty optimistic. For years, Elon Musk, CEO of Tesla and SpaceX, has been famous for his skeptical attitude towards AI and suggested it could be dangerous to the future of the human race.

SHANGHAI-BASED online English learning platform Liulishuo said yesterday it has raised US$100 million from institutional investors and previous investors to fuel its future growth into artificial intelligence and tailor-made programs for English learners. China Media Capital and Wu Capital, as well as previous investors including TrustBridge, IDG Capital, GGV Capital, Cherubic Ventures and Hearst Ventures, have been announced as investors in the platform. Wang Yi, co-founder and chief executive officer of Liulishuo, said the company plans to hire more talent in the artificial intelligence field and to offer more AI-driven educational services besides its current AI-powered personalized interactive courses. "We hope to maintain our leading position in the artificial intelligence-backed online education field and to further enhance efficiency in English learning," he said, adding that they also hope to build an artificial intelligence learning research institution within two or three years. It will also provide AI-backed spoken English evaluating services for NASDAQ-listed TAL Education Group to integrate with TAL's current learning systems.