Recent studies have shown that proximal gradient (PG) method and accelerated gradient method (APG) with restarting can enjoy a linear convergence under a weaker condition than strong convexity, namely a quadratic growth condition (QGC). However, the faster convergence of restarting APG method relies on the potentially unknown constant in QGC to appropriately restart APG, which restricts its applicability. We address this issue by developing a novel adaptive gradient converging methods, i.e., leveraging the magnitude of proximal gradient as a criterion for restart and termination. Our analysis extends to a much more general condition beyond the QGC, namely the H\"{o}lderian error bound (HEB) condition. {\it The key technique} for our development is a novel synthesis of {\it adaptive regularization and a conditional restarting scheme}, which extends previous work focusing on strongly convex problems to a much broader family of problems. Furthermore, we demonstrate that our results have important implication and applications in machine learning: (i) if the objective function is coercive and semi-algebraic, PG's convergence speed is essentially $o(\frac{1}{t})$, where $t$ is the total number of iterations; (ii) if the objective function consists of an $\ell_1$, $\ell_\infty$, $\ell_{1,\infty}$, or huber norm regularization and a convex smooth piecewise quadratic loss (e.g., squares loss, squared hinge loss and huber loss), the proposed algorithm is parameter-free and enjoys a {\it faster linear convergence} than PG without any other assumptions (e.g., restricted eigen-value condition). It is notable that our linear convergence results for the aforementioned problems are global instead of local. To the best of our knowledge, these improved results are the first shown in this work.

It has long been assumed that high dimensional continuous control problems cannot be solved effectively by discretizing individual dimensions of the action space due to the exponentially large number of bins over which policies would have to be learned. In this paper, we draw inspiration from the recent success of sequence-to-sequence models for structured prediction problems to develop policies over discretized spaces. Central to this method is the realization that complex functions over high dimensional spaces can be modeled by neural networks that use next step prediction. Specifically, we show how Q-values and policies over continuous spaces can be modeled using a next step prediction model over discretized dimensions. With this parameterization, it is possible to both leverage the compositional structure of action spaces during learning, as well as compute maxima over action spaces (approximately). On a simple example task we demonstrate empirically that our method can perform global search, which effectively gets around the local optimization issues that plague DDPG and NAF. We apply the technique to off-policy (Q-learning) methods and show that our method can achieve the state-of-the-art for off-policy methods on several continuous control tasks.

In this study an Artificial Neural Network was trained to classify musical instruments, using audio samples transformed to the frequency domain. Different features of the sound, in both time and frequency domain, were analyzed and compared in relation to how much information that could be derived from that limited data. The study concluded that in comparison with the base experiment, that had an accuracy of 93.5%, using the attack only resulted in 80.2% and the initial 100 Hz in 64.2%.

MindMeld has pioneered the development of a unique AI platform that enables customers to build intelligent and human-like conversational interfaces for any application or device. Through its proprietary machine learning (ML) technology, MindMeld delivers incredible accuracy to help users interact with voice and chat assistants in a more natural way. At Cisco, we believe that AI and machine learning will play an increasingly vital role across all parts of our business. Given our industry leadership in networking, collaboration, data center, analytics, and unified communications, we are uniquely positioned to take advantage of AI and ML and embed it across the network and the cloud. Cisco is unlocking capabilities through AI that were unthinkable in the past – for example, empowering our customers to self-manage their network and data center, stay ahead of security attacks, embed intelligence at the edge, deliver predictive analytics, and revolutionize the workplace.

Doctors are valued for our ability to use a combination of logic, memorised knowledge and communication skills to perform three main functions. The first is to diagnose a medical condition, the second is to treat it. The third, and most important function is to communicate information to another human being with empathy, in a way that's tailored to their ability to understand, and in a way which makes them feel cared for. With increasing demand for healthcare, and with stressed out doctors stretched ever thinner, the tendency is for doctors to concentrate on the diagnosing and treating. The emotional, concerned, human aspect of treating disease is neglected, but this function is the only one even the most advanced artificial intelligence is unlikely to ever genuinely perform. Medicine has long been considered an art, with "clinical acumen" the major skill of a doctor.

Welcome to the Hackett Group First Quarter Earnings Conference Call. Your lines have been placed on listed only mode. Hosting tonight's call are Mr. Ted Fernandez, Chairman and CEO; and Mr. Rob Ramirez, Chief Financial Officer. Mr. Ramirez, you may begin. Good afternoon, everyone, and thank you for joining us to discuss The Hackett Group's First Quarter Results. Speaking on the call today and here to answer your questions are Ted Fernandez, Chairman and CEO of The Hackett Group; and myself, Rob Ramirez, Chief Financial Officer. Our press announcement was released over the wires at 4:14 p.m. Eastern Time. For a copy of the release, please visit our website at www.thehackettgroup.com. We will also place any additional financial or statistical data discussed on this call that is not contained in the release on the Investor Relations page of our website. Before we begin, I would like to remind you that in the following comments and in the Q&A session, we will be making statements about expected future results, which may be forward-looking statements for the purposes of the federal securities laws. These statements relate to our current expectations, estimates and projections and are not a guarantee of future performance.

Mother's Day is coming up this weekend, but if you have an especially high IQ, you may want to get something for dad, too. A new study has found fathers have an important impact on their baby's cognitive development, and those whose fathers were positively active during the first few months of birth, scored higher on cognitive tests by age 2. The study, conducted at the Imperial College London, notes fathers who positively interact with their young children by playing with them and giving them positive feedback, help increase their offspring's cognitive development. In fact, the researchers suggest it's able to predict a young child's cognitive abilities by simply looking at how much good quality time they spent with dad. "The clear message for new fathers here is to get stuck in and play with your baby," said study author Professor Paul Ramchandani told The Independent. "Even when they're really young playing and interacting with them can have a positive effect."

Apple Watch might be more than a fancy accessory for your wrist. The device could be of great help to heart patients, according to a'Health e-heart' study conducted by University of California, San Francisco, which finds that the device is 97 percent accurate in diagnosing irregular heartbeat. "Our results show that common wearable trackers like smartwatches present a novel opportunity to monitor, capture and prompt medical therapy for atrial fibrillation without any active effort from patients. While mobile technology screening won't replace more conventional monitoring methods, it has the potential to successfully screen those at an increased risk and lower the number of undiagnosed cases of AF," the report's senior author, Gregory M. Marcus, MD, MAS Endowed Professor of Atrial Fibrillation Research and Director of Clinical Research for the Division of Cardiology at UCSF, said in the findings published Wednesday. The research trained a deep neural network (DNN) and paired it with the Apple Watch and Cardiogram app.

Cisco today announced plans to acquire AI startup MindMeld for $125 million. Cisco decided to acquire MindMeld, a company that's been called "Siri on steroids," because many bot experiences today are disappointing, IoT and applications VP Rowan Trollope wrote in a blog post today. "Bringing the MindMeld team to Cisco is a giant leap forward in helping our customers experience the next generation of interactive, conversational interfaces." Cisco collaboration software competition includes other enterprise chat players including Microsoft Teams, Google Hangouts Chat, and Slack.

NVIDIA has reported Q1 revenue of $1.94 billion, buoyed by record datacenter sales of $409 million. The datacenter business has been growing by leaps and bounds for the graphics chipmaker, thanks largely to the rapidly expanding market for high GPUs in deep learning. Revenue for the first quarter (Q1 FY18) beat analyst estimates and represents a 48 percent increase from last year at this time. Even more encouraging was net income, which amounted to $507 million for the quarter. That's more than twice that reported for the first quarter in 2016.