In this work, we derive a generic overcomplete frame thresholding scheme based on risk minimization. Overcomplete frames being favored for analysis tasks such as classification, regression or anomaly detection, we provide a way to leverage those optimal representations in real-world applications through the use of thresholding. We validate the method on a large scale bird activity detection task via the scattering network architecture performed by means of continuous wavelets, known for being an adequate dictionary in audio environments.

SAN ANTONIO--In a lower-for-longer environment, the implementation of artificial intelligence (AI) has come into play to cut costs in the oil and gas industry, according to Quantico Energy's CEO and founder, Barry Zhang.

The young woman in the video sits on a high bed in a comfortable, well-adorned bedroom. She touches the screen of her mobile phone to "see" a photo a friend posted on Facebook.

With its long-term accumulation in the areas of software-defined processors and heterogeneous computing, multi-core heterogeneous processor platform launched by Wuxi DSP Technologies has advanced performance power ratio with ease to the software development. The experience of multiple company partners shows that in a short period of time, a small-scale R & D team can quickly develop industry-leading innovations, dramatically shorten the product development cycle, reduce project investment risks and improve market response ability. With its innovative platform solutions, Wuxi DSP Technologies is committed to joining hands with numerous partners in the whole (upstream and downstream) industry chain to build the "Wuxi DSP" chip ecosystem by providing chip customization services with the goal of "openness, customization, shared-creation, security" and the strategy of "ecosystem co-construction, resource sharing, collaborative development, value enhancement". During the conference, partners such as China Mobile, Beijing Acoinfo Technology, Guangzhou Power Electrical Technology, Akceler Hardware Accelerator, Nationz Technologies shared their experiences in IoT eco-construction, technology and business model innovations respectively. Many vendors, including Shenzhen Micro & Nano Institute, Extra Dimension Technology, Hangzhou Langyang Technologies, Sun Kaiseng Technology, Beijing Sigbean Information Technology, Beijing Orihard Technology introduced their latest R & D achievements based on "Wuxi DSP" processor platform from IoT perception, computing, connectivity, AI cloud collaborative services, etc. Sample products were showcased onsite across multiple industries including communications, power sector, municipal administration and smart homes, which cover the hottest fields of IoT multimode communications, embedded real-time operating system, broadband power line carrier, microphone array, deep learning, multimode communications, millimeter wave radar, gesture recognition, smart metering, intelligent speaker and others.

Japanese researchers have developed a humanoid robot that is sportier than many human beings. He can play push-ups, pull-ups, badminton and even sweat. Because they really want me to feel like the lazy person I am, robots can now also do sit-ups, push-ups and something called back extensions. I had to look up the latter. Researchers from the University of Tokyo have described in the journal Science Robotics their humanoid, which looks amazingly lifelike in its appearance and movements.

For the new year, we asked 18 of our researchers and staff members from around PARC to share some of their technology predictions for 2018 and beyond. "Deep learning and deep reinforcement learning has revolutionized data analytics, but the amount of data required is still relatively huge. It's also not how humans learn – typically a few examples are enough to form a mental model. I think this is going to change as research shifts from feasibility and accuracy to scalability and transparency. There's already some work in this area of slow learning, including from PARC, and this will only get more prominent."

WASHINGTON: NASA has used Google's artificial intelligence (AI) to discover a record-tying eighth exoplanet circling a Sun-like star 2,545 light-years from Earth, marking the first finding of an eight-planet solar system like ours. Kepler-90i - a sizzling hot, rocky planet that orbits its star once every 14.4 days - was found using machine learning from Google to scour data from NASA's planet-hunting Kepler Telescope. "The Kepler-90 star system is like a mini version of our solar system. You have small planets inside and big planets outside, but everything is scrunched in much closer," said Andrew Vanderburg, a NASA Sagan Postdoctoral Fellow and astronomer at the University of Texas at Austin. Machine learning is an approach to artificial intelligence in which computers "learn."

Stories predicting that robots will one day take over your job are probably clogging your newsfeed and filling you with dread. But today's reality includes a lot of good news when it comes to AI: There are AI apps and tools available now that could make your job much easier and could improve your small business. Intuitive personal assistants, messengers, and data filters eliminate some of the grunt work that has been holding you back from advancing your organization - and your career. At CreditDonkey, we compare the latest products and services for companies so that they can decide which ones are right for them - and spend wisely on their choice. With the latest AI innovations, companies may find they can save money in the long run by working more efficiently and getting closer to the information they need to run more smoothly. With so many new AI products, though, you can fall down a rabbit hole of options - and not know which ones would be worth investing in. The AI innovations on our list are worth considering; they're the best of the best. See which AI tools on this list could be a fit for your business. Jibo is an AI-powered robot that learns something new with every experience, recognizing faces, telling jokes, and more.

Random backpropagation (RBP) is a variant of the backpropagation algorithm for training neural networks, where the transpose of the forward matrices are replaced by fixed random matrices in the calculation of the weight updates. It is remarkable both because of its effectiveness, in spite of using random matrices to communicate error information, and because it completely removes the taxing requirement of maintaining symmetric weights in a physical neural system. To better understand random backpropagation, we first connect it to the notions of local learning and learning channels. Through this connection, we derive several alternatives to RBP, including skipped RBP (SRPB), adaptive RBP (ARBP), sparse RBP, and their combinations (e.g. ASRBP) and analyze their computational complexity. We then study their behavior through simulations using the MNIST and CIFAR-10 bechnmark datasets. These simulations show that most of these variants work robustly, almost as well as backpropagation, and that multiplication by the derivatives of the activation functions is important. As a follow-up, we study also the low-end of the number of bits required to communicate error information over the learning channel. We then provide partial intuitive explanations for some of the remarkable properties of RBP and its variations. Finally, we prove several mathematical results, including the convergence to fixed points of linear chains of arbitrary length, the convergence to fixed points of linear autoencoders with decorrelated data, the long-term existence of solutions for linear systems with a single hidden layer and convergence in special cases, and the convergence to fixed points of non-linear chains, when the derivative of the activation functions is included.

If 2017 left you breathless, exhausted by unexpected headlines, then brace yourself. The coming year may bring even more turbulent change, according to the CEOs, academics, economists and other bold thinkers we consulted for our annual peek at the year ahead.