Independent Component Analysis (ICA) is a technique for unsupervised exploration of multi-channel data that is widely used in observational sciences. In its classic form, ICA relies on modeling the data as linear mixtures of non-Gaussian independent sources. The maximization of the corresponding likelihood is a challenging problem if it has to be completed quickly and accurately on large sets of real data. We introduce the Preconditioned ICA for Real Data (Picard) algorithm, which is a relative L-BFGS algorithm preconditioned with sparse Hessian approximations. Extensive numerical comparisons to several algorithms of the same class demonstrate the superior performance of the proposed technique, especially on real data, for which the ICA model does not necessarily hold.

This paper introduces a novel approach to texture synthesis based on generative adversarial networks (GAN) (Goodfellow et al., 2014). We extend the structure of the input noise distribution by constructing tensors with different types of dimensions. We call this technique Periodic Spatial GAN (PSGAN). The PSGAN has several novel abilities which surpass the current state of the art in texture synthesis. First, we can learn multiple textures from datasets of one or more complex large images. Second, we show that the image generation with PSGANs has properties of a texture manifold: we can smoothly interpolate between samples in the structured noise space and generate novel samples, which lie perceptually between the textures of the original dataset. In addition, we can also accurately learn periodical textures. We make multiple experiments which show that PSGANs can flexibly handle diverse texture and image data sources. Our method is highly scalable and it can generate output images of arbitrary large size.

Generative adversarial networks (GANs) [7] are a recent approach to train generative models of data, which have been shown to work particularly well on image data. In the current paper we introduce a new model for texture synthesis based on GAN learning. By extending the input noise distribution space from a single vector to a whole spatial tensor, we create an architecture with properties well suited to the task of texture synthesis, which we call spatial GAN (SGAN). To our knowledge, this is the first successful completely data-driven texture synthesis method based on GANs. Our method has the following features which make it a state of the art algorithm for texture synthesis: high image quality of the generated textures, very high scalability w.r.t. the output texture size, fast real-time forward generation, the ability to fuse multiple diverse source images in complex textures. To illustrate these capabilities we present multiple experiments with different classes of texture images and use cases. We also discuss some limitations of our method with respect to the types of texture images it can synthesize, and compare it to other neural techniques for texture generation.

Convolutional sparse representations are a form of sparse representation with a dictionary that has a structure that is equivalent to convolution with a set of linear filters. While effective algorithms have recently been developed for the convolutional sparse coding problem, the corresponding dictionary learning problem is substantially more challenging. Furthermore, although a number of different approaches have been proposed, the absence of thorough comparisons between them makes it difficult to determine which of them represents the current state of the art. The present work both addresses this deficiency and proposes some new approaches that outperform existing ones in certain contexts. A thorough set of performance comparisons indicates a very wide range of performance differences among the existing and proposed methods, and clearly identifies those that are the most effective.

Fully convolutional neural networks (FCN) have been shown to achieve state-of-the-art performance on the task of classifying time series sequences. We propose the augmentation of fully convolutional networks with long short term memory recurrent neural network (LSTM RNN) sub-modules for time series classification. Our proposed models significantly enhance the performance of fully convolutional networks with a nominal increase in model size and require minimal preprocessing of the dataset. The proposed Long Short Term Memory Fully Convolutional Network (LSTM-FCN) achieves state-of-the-art performance compared to others. We also explore the usage of attention mechanism to improve time series classification with the Attention Long Short Term Memory Fully Convolutional Network (ALSTM-FCN). Utilization of the attention mechanism allows one to visualize the decision process of the LSTM cell. Furthermore, we propose fine-tuning as a method to enhance the performance of trained models. An overall analysis of the performance of our model is provided and compared to other techniques.

IBM is officially partnering with the Massachusetts Institute of Technology (MIT) to run an artificial intelligence (AI) research lab. This Watson-branded joint MIT-IBM AI research initiative -- a partnership of acronyms -- will be funded through a 10-year, $240 million investment from IBM and will be co-located at IBM's Research Lab in Cambridge and at the MIT campus. According to an IBM press release, the MIT-IBM Watson AI Lab will be one the largest long-term AI collaborations between a university and a member of the tech industry. Its goal is to enable more than 100 scientists, professors, and students to pursue research focused on such areas as the development of AI algorithms that could expand machine learning capabilities, the improvement of AI hardware, the exploration of AI's economic and societal benefits, and the identification of AI applications in key industries. The team at IBM knows that despite all the recent attention given to what AI can do, there's still so much that it can't.

Artificial intelligence can accurately predict whether people are gay or straight based on photos of their faces, according to new research suggesting that machines can have significantly better "gaydar" than humans. The study from Stanford University – which found that a computer algorithm could correctly distinguish between gay and straight men 81% of the time, and 74% for women – has raised questions about the biological origins of sexual orientation, the ethics of facial-detection technology and the potential for this kind of software to violate people's privacy or be abused for anti-LGBT purposes. The machine intelligence tested in the research, which was published in the Journal of Personality and Social Psychology and first reported in the Economist, was based on a sample of more than 35,000 facial images that men and women publicly posted on a US dating website. The researchers, Michal Kosinski and Yilun Wang, extracted features from the images using "deep neural networks", meaning a sophisticated mathematical system that learns to analyze visuals based on a large dataset. The research found that gay men and women tended to have "gender-atypical" features, expressions and "grooming styles", essentially meaning gay men appeared more feminine and visa versa.

About a month ago, iRobot CEO Colin Angle mentioned something about sharing Roomba mapping data in an interview with Reuters. It got turned into a data privacy kerfuffle in a way that iRobot did not intend and (probably) did not deserve, as evidenced by their immediate clarification that iRobot will not sell your data or share it without your consent. Data privacy is important, of course, especially for devices that live in your home with you. But as robots get more capable, the amount of data that they collect will increase, and sharing that data in a useful, thoughtful, and considerate way could make smart homes way smarter. To understand how iRobot is going to make this happen, we spoke with Angle about keeping your data safe, integrating robots with the future smart home, and robots that can get you a beer.

Over-fitting.If you perform a regression with 200 predictors (with strong cross-correlations among predictors), use meta regression coefficients: that is, use coefficients of the form f[Corr(Var, Response), a,b, c] where a, b, c are three meta-parameters (e.g. This will reduce your number of parameters from 200 to 3, and eliminate most of the over-fitting Perform the right type of cross-validation. If your training set has 400,000 observations distributed across 50 clients, and your test data set (used for cross-validation) has 200,000 observations but only 3 clients or 5 days worth of historical data, then your cross-validation methodology is very flawed. Better, split your cross-validation data set in 5 subsets to compute confidence intervals. Make sure you've eliminated outliers and cleaned your data set.

The future of the U.S. military may be focused on artificial intelligence (A.I.), an effort that could improve cybersecurity, precision weaponry and other military functions. A recent report from the Harvard's Belfer Center for Science and International Affairs states that advancements in the last five years have made it possible for the U.S. military to expand its use of A.I. in the near future-- but only if certain questions are addressed first. "Though the United States military and intelligence communities are planning for expanded use of A.I. across their portfolios, many of the most transformative applications of A.I. have not yet been addressed," the report, written by Greg Allen and Taniel Chan, states. "We propose three goals for developing future policy on A.I. and national security: preserving U.S. technological leadership, supporting peaceful and commercial use and mitigating catastrophic risk." The researchers examined nuclear, aerospace, cyber and biotech opportunities to develop recommendations for national security policy involving A.I .funding