Deep learning models' architectures, including depth and width, are key factors influencing models' performance, such as test accuracy and computation time. This paper solves two problems: given computation time budget, choose an architecture to maximize accuracy, and given accuracy requirement, choose an architecture to minimize computation time. We convert this architecture optimization into a subset selection problem. With accuracy's submodularity and computation time's supermodularity, we propose efficient greedy optimization algorithms. The experiments demonstrate our algorithm's ability to find more accurate models or faster models. By analyzing architecture evolution with growing time budget, we discuss relationships among accuracy, time and architecture, and give suggestions on neural network architecture design.

Recurrent neural network (RNN)'s architecture is a key factor influencing its performance. We propose algorithms to optimize hidden sizes under running time constraint. We convert the discrete optimization into a subset selection problem. By novel transformations, the objective function becomes submodular and constraint becomes supermodular. A greedy algorithm with bounds is suggested to solve the transformed problem. And we show how transformations influence the bounds. To speed up optimization, surrogate functions are proposed which balance exploration and exploitation. Experiments show that our algorithms can find more accurate models or faster models than manually tuned state-of-the-art and random search. We also compare popular RNN architectures using our algorithms.

Variational methods that rely on a recognition network to approximate the posterior of directed graphical models offer better inference and learning than previous methods. Recent advances that exploit the capacity and flexibility in this approach have expanded what kinds of models can be trained. However, as a proposal for the posterior, the capacity of the recognition network is limited, which can constrain the representational power of the generative model and increase the variance of Monte Carlo estimates. To address these issues, we introduce an iterative refinement procedure for improving the approximate posterior of the recognition network and show that training with the refined posterior is competitive with state-of-the-art methods. The advantages of refinement are further evident in an increased effective sample size, which implies a lower variance of gradient estimates.

Understanding recent evolution of object detection and localization with intuitive explanation of underlying concepts. Object detection is one of the areas of computed vision that is maturing very rapidly. Every year, new algorithms/ models keep on outperforming the previous ones. In-fact, one of the latest state of the art software system for object detection was just released last week by Facebook AI team. The software is called Detectron that incorporates numerous research projects for object detection and is powered by the Caffe2 deep learning framework.

Intraspexion uses deep learning and predictive analytics to predict and prevent potential litigation. Intraspexion is as an early warning system that operates through analysis of a company's emails to identify those that contain risk factors. Legal lawsuits cost companies in the USA $150 billion each year. Deep learning is used to identify specific, potential risks to an enterprise (of which litigation is the prime example) while such risks are still internal electronic communications. The system involves mining and using existing classifications of data (e.g., from a litigation database) to train one or more deep learning algorithms, and then examining the internal electronic communications with the trained algorithm, to generate a scored output that will enable enterprise personnel to be alerted to risks and take action in time to prevent the risks from resulting in harm to the enterprise or others.

The inaugural Deep Learning World (all about the commercial deployment of deep learning) is headed to Las Vegas alongside Predictive Analytics World and we're excited to announce a top-notch list of speakers and companies who are confirmed to deliver sessions at Caesar's Palace on June 3-7, 2018. Don't miss talks from Capital One, Microsoft, PayPal, and more. Get started at the full-day, hands-on workshop "Deep Learning in Practice" Speakers & Session Topics Include: Check out the speaker list and stay tuned for more announcements. Register now and save with Early Bird Prices to the first ever Deep Learning World conference. In 2018, there will be only ONE PAW in the U.S. – Mega-PAW – with five (5) parallel events amounting to seven (7) tracks: PAW Business, PAW Financial, PAW Healthcare, PAW Manufacturing, and Deep Learning World.

Artificial Intelligence (AI) is hot. Although its capabilities have been steadily increasing for years, it was the victory of DeepMind's AlphaGo program over the top Go expert Lee Se-dol last year that alerted many to the rapid pace of development in the AI field.

These authors contributed equally: Ryan Poplin, Avinash V. Varadarajan, Lily Peng and Dale R. Webster. The authors are employees of Google and Verily Life Sciences.

In a paper (PDF) published today in the Nature journal Biomedical Engineering, researchers explained their method: An AI algorithm evaluated eye scans and, after refining its model with machine learning, was able to predict cardiovascular risk factors like age, gender and blood pressure. This could lead to easier and potentially quicker analysis than a blood test with roughly the same accuracy as current methods. The study isn't without limitations, given that it only surveyed eye images with a 45-degree field of view. More research would resolve whether the model needs to be adjusted for larger or smaller photos, and a larger data set than what the researchers used is more appropriate for deep learning.

Interest in machine learning has exploded over the past decade. You see machine learning in computer science programs, industry conferences, and the Wall Street Journal almost daily. For all the talk about machine learning, many conflate what it can do with what they wish it could do. Fundamentally, machine learning is using algorithms to extract information from raw data and represent it in some type of model. We use this model to infer things about other data we have not yet modeled. Neural networks are one type of model for machine learning; they have been around for at least 50 years.