To advance the functionality of today's home appliances to a whole new level, LG Electronics (LG) is set to deliver an unparalleled level of performance and convenience into the home with deep learning technology to be unveiled at CES 2017. LG deep learning will allow home appliances to better understand their users by gathering and studying customers' lifestyle patterns over time. This process never ends and improves over time to provide customers with new solutions to everyday problems. Using multiple sensors and LG's deep learning technology, LG's newest robot vacuum cleaner will recognize objects around the room and react accordingly. By capturing surface images of the room, the intelligent cleaner remembers obstacles and learns to avoid them over time.

Our Head of Research, Alec Radford, recently led a workshop on general sequence learning using recurrent neural networks at Next.ML in San Francisco. Next.ML was created to teach the latest actionable machine learning techniques that you can use right out of the workshop.The upcoming Next.ML workshop will be in Cambridge, MA at the Microsoft NERD Center on April 27. Recurrent Neural Networks hold great promise as general sequence learning algorithms. As such, they are a very promising tool for text analysis. However, outside of very specific use cases such as handwriting recognition and recently, machine translation, they have not seen widespread use.

A computer's ability to accurately identify images is a white whale for many technology companies, from Baidu to Google. One Australian startup has found a corner of the market to dominate, winning contracts with the European Union Intellectual Property Office (EUIPO) and IP Australia for algorithms that can detect and compare logos. SEE ALSO: Airbnb is getting into the airline booking disruption game with'Flights' TrademarkVision, which has support from Australia's CEA Startup Fund, uses machine learning to support image searches that can identify similar trademarks. Having a unique trademark or logo is vital, but many intellectual property registration bodies often require outdated forms of non-visual search that make comparison difficult. Australia, for example, relies on keywords, Europe on Vienna codes and the U.S. on design codes.

When we offered CS231n (Deep Learning class) at Stanford, we intentionally designed the programming assignments to include explicit calculations involved in backpropagation on the lowest level. The students had to implement the forward and the backward pass of each layer in raw numpy. This is seemingly a perfectly sensible appeal - if you're never going to write backward passes once the class is over, why practice writing them? Are we just torturing the students for our own amusement? Some easy answers could make arguments along the lines of "it's worth knowing what's under the hood as an intellectual curiosity", or perhaps "you might want to improve on the core algorithm later", but there is a much stronger and practical argument, which I wanted to devote a whole post to: In other words, it is easy to fall into the trap of abstracting away the learning process -- believing that you can simply stack arbitrary layers together and backprop will "magically make them work" on your data.

Financial Portfolio theories are one of the important achievements in financial economics in the last XX Century. One such theory goes by the designation of Markowitz Portfolio Theory or Modern Portfolio Theory, named after Nobel Prize in Economic Sciences winner Harry Markowitz. We read the Wikipedia entry for this theory and we can immediately confirm it as a mathematical and statistical theory at its core. And if it is mathematical and statistical at its core it is well positioned to be improved and enhanced by an algorithmic, computational approach. And that is the case with our paper's proposal: it is another one software approach to Portfolio Theory that turns the problem of finding the best efficient frontier predicted by the theory into a mathematical optimization problem, but from the new machine learning/deep learning perspective.

London artificial intelligence lab DeepMind is setting up a sizeable new team in the US in a bid to increase collaboration with parent company Google. DeepMind, bought by Google in 2014 for £400 million, is planning to hire "a couple of dozen" people at Google's headquarters in Mountain View, according to a DeepMind spokesperson. "We're proud to already have close partnerships with many teams at Google, but we're yet to develop an algorithm that gets rid of time zone differences," the spokesperson told Business Insider. "So we're hiring a small DeepMind Applied team in Mountain View to bridge the gap between Google and our team in London, helping us collaborate even more closely to bring our research breakthroughs to Google users around the world." The expansion represents a significant milestone in DeepMind's journey and comes after Yann LeCun, the head of AI research at Facebook, suggested that DeepMind was too far away from the Google "mothership" to have a significant impact.

We hear and read in the popular media about Artificial Intelligence (AI) all the time. We have movies about them. We hear about Elon Musk and Stephen Hawking warning us about AI's apocalyptic consequences. We hear from the World Economics forum about AI's effect on taking away our jobs. We here about how disruptive AI will be for businesses.

This paper introduces Graph Convolutional Recurrent Network (GCRN), a deep learning model able to predict structured sequences of data. Precisely, GCRN is a generalization of classical recurrent neural networks (RNN) to data structured by an arbitrary graph. Such structured sequences can represent series of frames in videos, spatio-temporal measurements on a network of sensors, or random walks on a vocabulary graph for natural language modeling. The proposed model combines convolutional neural networks (CNN) on graphs to identify spatial structures and RNN to find dynamic patterns. We study two possible architectures of GCRN, and apply the models to two practical problems: predicting moving MNIST data, and modeling natural language with the Penn Treebank dataset. Experiments show that exploiting simultaneously graph spatial and dynamic information about data can improve both precision and learning speed.

Currently there are two predominant ways to train deep neural networks. The first one uses restricted Boltzmann machine (RBM) and the second one autoencoders. RBMs are stacked in layers to form deep belief network (DBN); the final representation layer is attached to the target to complete the deep neural network. Autoencoders are nested one inside the other to form stacked autoencoders; once the stcaked autoencoder is learnt the decoder portion is detached and the target attached to the deepest layer of the encoder to form the deep neural network. This work proposes a new approach to train deep neural networks using dictionary learning as the basic building block; the idea is to use the features from the shallower layer as inputs for training the next deeper layer. One can use any type of dictionary learning (unsupervised, supervised, discriminative etc.) as basic units till the pre-final layer. In the final layer one needs to use the label consistent dictionary learning formulation for classification. We compare our proposed framework with existing state-of-the-art deep learning techniques on benchmark problems; we are always within the top 10 results. In actual problems of age and gender classification, we are better than the best known techniques.

The computational properties of neural systems are often thought to be implemented in terms of their network dynamics. Hence, recovering the system dynamics from experimentally observed neuronal time series, like multiple single-unit (MSU) recordings or neuroimaging data, is an important step toward understanding its computations. Ideally, one would not only seek a state space representation of the dynamics, but would wish to have access to its governing equations for in-depth analysis. Recurrent neural networks (RNNs) are a computationally powerful and dynamically universal formal framework which has been extensively studied from both the computational and the dynamical systems perspective. Here we develop a semi-analytical maximum-likelihood estimation scheme for piecewise-linear RNNs (PLRNNs) within the statistical framework of state space models, which accounts for noise in both the underlying latent dynamics and the observation process. The Expectation-Maximization algorithm is used to infer the latent state distribution, through a global Laplace approximation, and the PLRNN parameters iteratively. After validating the procedure on toy examples, the approach is applied to MSU recordings from the rodent anterior cingulate cortex obtained during performance of a classical working memory task, delayed alternation. A model with 5 states turned out to be sufficient to capture the essential computational dynamics underlying task performance, including stimulus-selective delay activity. The estimated models were rarely multi-stable, but rather were tuned to exhibit slow dynamics in the vicinity of a bifurcation point. In summary, the present work advances a semi-analytical (thus reasonably fast) maximum-likelihood estimation framework for PLRNNs that may enable to recover the relevant dynamics underlying observed neuronal time series, and directly link them to computational properties.