What's better than an app that turns selfies into instant art? An app that does the same thing for video and virtual reality. That's what Prisma, the app that has already captured the imagination of the photo-sharing community, is working to bring to the public next. Using a combination of neural networks and artificial intelligence, the app takes normal, real world images and transforms them into approximations of how painters might render them into famous art styles (abstract, impressionist, etc.). After contacting the Russia-based developers behind the app, we confirmed, as noted on Sunday, that the app will soon move into video.

How often do you ring your bank and forget the special dates, places or names needed just to get through security? Apparently, it takes us 45 seconds on average just to confirm who we are. Now that might not seem very long, but if you're a global bank like Citi receiving 35 million calls a year in Asia alone, that adds up to 437,500 hours of staff time a year. But by using computers to identify our voices, this authentication process can be cut to 15 seconds on average, saving the bank pots of cash and us lots of hassle. Citi has just begun rolling out this kind of voice biometrics authentication for its 15 million Asian banking customers, starting in Taiwan, Australia, Hong Kong and Singapore.

The Walt Disney Co. is kicking off the third session of its corporate accelerator this week, and revealed 9 new companies admitted to the program. A full list follows at the end of this post. The companies are developing everything from cinematic virtual reality and holographic content, to robots with human-like facial expressions. Because alumni of the Disney Accelerator have scored big partnerships with the media and entertainment juggernaut in the past, it is seen as one of the more desirable corporate accelerators out there. According to research by Future Asia Ventures, there are 131 active corporate accelerators worldwide today, with 13 new programs launching in the first half of 2016.

The well known maximum-entropy principle due to Jaynes, which states that given mean parameters, the maximum entropy distribution matching them is in an exponential family, has been very popular in machine learning due to its "Occam's razor" interpretation. Unfortunately, calculating the potentials in the maximum-entropy distribution is intractable \cite{bresler2014hardness}. We provide computationally efficient versions of this principle when the mean parameters are pairwise moments: we design distributions that approximately match given pairwise moments, while having entropy which is comparable to the maximum entropy distribution matching those moments. We additionally provide surprising applications of the approximate maximum entropy principle to designing provable variational methods for partition function calculations for Ising models without any assumptions on the potentials of the model. More precisely, we show that in every temperature, we can get approximation guarantees for the log-partition function comparable to those in the low-temperature limit, which is the setting of optimization of quadratic forms over the hypercube. \cite{alon2006approximating}

High-dimensional big data appears in many research fields such as image recognition, biology and collaborative filtering. Often, the exploration of such data by classic algorithms is encountered with difficulties due to `curse of dimensionality' phenomenon. Therefore, dimensionality reduction methods are applied to the data prior to its analysis. Many of these methods are based on principal components analysis, which is statistically driven, namely they map the data into a low-dimension subspace that preserves significant statistical properties of the high-dimensional data. As a consequence, such methods do not directly address the geometry of the data, reflected by the mutual distances between multidimensional data point. Thus, operations such as classification, anomaly detection or other machine learning tasks may be affected. This work provides a dictionary-based framework for geometrically driven data analysis that includes dimensionality reduction, out-of-sample extension and anomaly detection. It embeds high-dimensional data in a low-dimensional subspace. This embedding preserves the original high-dimensional geometry of the data up to a user-defined distortion rate. In addition, it identifies a subset of landmark data points that constitute a dictionary for the analyzed dataset. The dictionary enables to have a natural extension of the low-dimensional embedding to out-of-sample data points, which gives rise to a distortion-based criterion for anomaly detection. The suggested method is demonstrated on synthetic and real-world datasets and achieves good results for classification, anomaly detection and out-of-sample tasks.

The Walt Disney Co. is kicking off the third session of its corporate accelerator this week, and revealed 9 new companies admitted to the program. A full list follows at the end of this post. The companies are developing everything from cinematic virtual reality and holographic content, to robots with human-like facial expressions. Because alumni of the Disney Accelerator have scored big partnerships with the media and entertainment juggernaut in the past, it is seen as one of the more desirable corporate accelerators out there. According to research by Future Asia Ventures, there are 131 active corporate accelerators worldwide today, with 13 new programs launching in the first half of 2016.

Financial terms of the deal were not disclosed. The acquisition is eBay's latest move to build up its machine learning capabilities in order to clean up its vast product catalog. "For our buyers, it will help us better understand the price differentiating attributes of our products, and, for our sellers, it will help us build out the predictive models that can define the probability of selling a given product, at a given price over time," explained Amit Menipaz, eBay's vice president and general manager of structured data. The acquisition of SalesPredict follows eBay's announcement in May that it is buying the Swedish machine-learning startup Expertmaker. Once the acquisition is complete, SalesPredict CEO Yaron Zakai-Or will serve as a director of product management, technology.

Priyadharshini writes on Project Management, IT, Six Sigma, & e-Learning. With a penchant for writing and a passion for professional education & development, she is adept at penning educative articles. She was previously associated with Oxford University Press and Pearson Education, India.

Today eBay announced it will acquire SalesPredict, an Israel-based company that leverages advanced analytics to predict customer buying behavior and sales conversion. SalesPredict is eBay's latest acquisition that will support its artificial intelligence, machine learning and data science efforts. It follows eBay's recent acquisition of Expertmaker, in order to further bolster our structured data efforts. Financial terms of the deal were not disclosed. Upon the close of the transaction, a number of SalesPredict's employees will join eBay's structured data organization, working from eBay's Israeli Development Center in Netanya.

A new artificial intelligence (AI) accelerator founded by a Hong Kong-born venture capitalist wants to bring Asia's best AI start-ups to the global stage. Dubbed Zeroth.ai, the AI accelerator was founded by Tak Lo, who most recently was a venture partner at Hong Kong venture capital company Mind Fund and a director at UK-based VC Techstars. "AI for me is something that will move the meter on technology in the next five to 10 years, there's no dispute about that," Lo said. The accelerator programme, which is currently accepting applications from AI start-ups in Asia, will officially begin in November this year. "I want to bring the wealth of experience and network from being a venture capitalist in New York and London to Asia and Hong Kong," Lo said.