Reinforcement learning and planning methods require an objective or reward function that encodes the desired behavior. Yet, in practice, there is a wide range of scenarios where an objective is difficult to provide programmatically, such as tasks with visual observations involving unknown object positions or deformable objects. In these cases, prior methods use engineered problem-specific solutions, e.g., by instrumenting the environment with additional sensors to measure a proxy for the objective. Such solutions require a significant engineering effort on a per-task basis, and make it impractical for robots to continuously learn complex skills outside of laboratory settings. We aim to find a more general and scalable solution for specifying goals for robot learning in unconstrained environments. To that end, we formulate the few-shot objective learning problem, where the goal is to learn a task objective from only a few example images of successful end states for that task. We propose a simple solution to this problem: meta-learn a classifier that can recognize new goals from a few examples. We show how this approach can be used with both model-free reinforcement learning and visual model-based planning and show results in three domains: rope manipulation from images in simulation, visual navigation in a simulated 3D environment, and object arrangement into user-specified configurations on a real robot.

The ability of modeling the other agents, such as understanding their intentions and skills, is essential to an agent's interactions with other agents. Conventional agent modeling relies on passive observation from demonstrations. In this work, we propose an interactive agent modeling scheme enabled by encouraging an agent to learn to probe. In particular, the probing agent (i.e. a learner) learns to interact with the environment and with a target agent (i.e., a demonstrator) to maximize the change in the observed behaviors of that agent. Through probing, rich behaviors can be observed and are used for enhancing the agent modeling to learn a more accurate mind model of the target agent. Our framework consists of two learning processes: i) imitation learning for an approximated agent model and ii) pure curiosity-driven reinforcement learning for an efficient probing policy to discover new behaviors that otherwise can not be observed. We have validated our approach in four different tasks. The experimental results suggest that the agent model learned by our approach i) generalizes better in novel scenarios than the ones learned by passive observation, random probing, and other curiosity-driven approaches do, and ii) can be used for enhancing performance in multiple applications including distilling optimal planning to a policy net, collaboration, and competition. A video demo is available at https://www.dropbox.com/s/8mz6rd3349tso67/Probing_Demo.mov?dl=0

Gather student data, make predictions about their learning--and perhaps their future. For years education companies have tried to apply technologies to better understand students and tailor their learning experiences, or support instructors who can intervene when human help is needed. Today the latest buzz revolves around machine learning, which education technologists claim can support more precise tools. And what it takes to make these products effective, and how to boost student learning equitably and ethically, remains an ongoing debate. Speakers quickly contextualized the technology with the shift in how widely available data is today.

The next article will be about Sequence modeling with Neural Networks. We will learn how to model sequences with a focus on Recurrent Neural Networks (RNNs) and their short-term memory and Long Short Term Memory (LSTM) and their ability to keep track of information throughout many timesteps. Bio: Zied Haj-Yahia is Senior Data Scientist at Capgemini Consulting. He specializes in building predictive models utilizing both traditional statistical methods and modern machine learning techniques. He also runs some workshops for university students (ESSEC, HEC, Ecole polytechnique) interested in Data Science and its applications. He is the co-founder of Global International Trading (GIT), a central purchasing office based in Paris.

This week Athena Security announces the successful launch and implementation of the world's first artificial intelligence security camera system capable of instantly and accurately recognizing an active shooter before they shoot, alerting law enforcement and verbally alerting the assailant that Police are en route. Historically, security cameras are only as useful as the people actively monitoring them. Without real time oversight, unattended security camera feeds only help to piece together crimes after the fact. Past AI and computer vision technologies provided far too many false positives and therefore could not be deployed because they would erroneously alert law enforcement every couple of minutes rendering them useless. Athena Security's AI-powered system provides authorities with real time video footage helping speed police and medical aid to any type of crime scene decreasing fatalities with faster response times.

At Rensselaer Polytechnic Institute, students are immersing themselves in Chinese culture without setting foot outside their classroom. The Mandarin Project, a collaboration between RPI, located in upstate New York, and the tech giant IBM, places students in a virtual world where they can practice their Mandarin language skills in a series of simulated scenarios, such as ordering lunch in a restaurant or taking a tai chi class. The project aims to make students feel as if they are actually in China, without the inconvenience of traveling there, says Helen Zhou, assistant professor of communication and media at RPI, who has been actively involved in designing the project. In a high-tech "cognitive immersive room," a classroom with a 360-degree floor-to-ceiling screen, students can practice their Mandarin with artificial intelligence-powered animated characters (including a floating panda head). The CIR combines several emerging technologies -- natural language processing, speech-to-text and movement tracking -- to create a unique learning experience, said Zhou.

"Don't worry, human intelligence will never be replaced by machines." That's what I was told as a freshman foreign languages student at a university. That was the time the concerns about the machine translation taking over the human, first came up. For an honest average playgoer, language is nothing but a set of words put in a specific order based on some (not so) simple rules. Learning languages is a grind.

If I could go back in time, I would try myself in Machine Learning 12 years ago! Right when I finished undergrad and came to the USA. After starting Andrew Ng's Machine Learning course on Coursera last month, I dropped everything except most urgent things and completed an 11 week course in just 3 weeks. The somewhat sad truth is, I first enrolled in this course many months ago, but I didn't start it then. Stars finally aligned in August and I started that course.

Artificial intelligence could have a profound impact on learning, but it also raises key questions. Artificial intelligence (AI) and machine learning are no longer fantastical prospects seen only in science fiction. Products like Amazon Echo and Siri have brought AI into many homes, and experts say it's only a matter of time before the technology has a profound impact in education, as well. Already, there are interactive tutors and adaptive learning programs that use AI to personalize instruction for students, and AI is also helping to simplify some administrative tasks. But Kelly Calhoun Williams, an education analyst for the technology research firm Gartner Inc., cautions there is a clear gap between the promise of AI and the reality of AI. "That's to be expected, given the complexity of the technology," she said.

Computer science professor Jacob Whitehill and his colleagues have received a $750,000, three-year grant from the National Science Foundation to develop the Automatic Classroom Observation Recognition Neural Network platform, or ACORN for short, which will combine machine learning, natural language processing, and elements of psychology and educational theory to deliver rapid feedback on teacher-student interactions. "I've always loved teaching – I really enjoy the dynamics between myself and my students," Mr. Whitehill said, adding that he's been particularly fascinated with finding a way to identify the characteristics of a positive exchange between instructor and pupil. Currently, teachers rely on one-on-one and group feedback, as well as some video-aided assessment, to evaluate those interactions. Mr. Whitehill believes ACORN could allow them to critique themselves not only much more frequently, but also without the pressure of a peer or supervisor being involved. "This kind of technology allows them to get objective feedback on their own teaching every day if they wanted to," he said.