This paper describes a novel energy-based probabilistic distribution that represents complex-valued data and explains how to apply it to direct feature extraction from complex-valued spectra. The proposed model, the complex-valued restricted Boltzmann machine (CRBM), is designed to deal with complex-valued visible units as an extension of the well-known restricted Boltzmann machine (RBM). Like the RBM, the CRBM learns the relationships between visible and hidden units without having connections between units in the same layer, which dramatically improves training efficiency by using Gibbs sampling or contrastive divergence (CD). Another important characteristic is that the CRBM also has connections between real and imaginary parts of each of the complex-valued visible units that help represent the data distribution in the complex domain. In speech signal processing, classification and generation features are often based on amplitude spectra (e.g., MFCC, cepstra, and mel-cepstra) even if they are calculated from complex spectra, and they ignore phase information. In contrast, the proposed feature extractor using the CRBM directly encodes the complex spectra (or another complex-valued representation of the complex spectra) into binary-valued latent features (hidden units). Since the visible-hidden connections are undirected, we can also recover (decode) the complex spectra from the latent features directly. Our speech coding experiments demonstrated that the CRBM outperformed other speech coding methods, such as methods using the conventional RBM, the mel-log spectrum approximate (MLSA) decoder, etc.

It has been observed that deep learning architectures tend to make erroneous decisions with high reliability for particularly designed adversarial instances. In this work, we show that the perturbation analysis of these architectures provides a framework for generating adversarial instances by convex programming which, for classification tasks, is able to recover variants of existing non-adaptive adversarial methods. The proposed framework can be used for the design of adversarial noise under various desirable constraints and different types of networks. Moreover, this framework is capable of explaining various existing adversarial methods and can be used to derive new algorithms as well. Furthermore, we make use of these results to obtain novel algorithms. Experiments show the competitive performance of the obtained solutions, in terms of fooling ratio, when benchmarked with well-known adversarial methods.

It might not look that special, but the robot above is, according to a new measure, the most dexterous one ever created. Among other tricks, it could sort through your junk drawer with unrivaled speed and skill. The key to its dexterity is not in its mechanical grippers but in its brain. The robot uses software called Dex-Net to determine how to pick up even odd-looking objects with incredible efficiency. The new robot was built by Ken Goldberg, a professor at UC Berkeley, and one of his graduate students, Jeff Mahler. Goldberg will demonstrate the latest version of it at EmTech Digital, an event in San Francisco organized by MIT Technology Review and dedicated to artificial intelligence.

Deep learning is well known for solving seemingly intractable problems in computer vision and natural language processing, but it typically does so by using massive CPU and GPU resources. Traditional deep learning techniques aren't well suited to addressing the challenges of Internet of Things (IoT) applications, however, because they can't apply the same level of computational resources. When running deep learning analysis on mobile devices, developers must adapt to a more resource-constrained platform. Image analysis on resource-constrained platforms can consume significant compute and memory resources. For example, the SpotGarbage app uses convolutional neural networks to detect garbage in images but consumes 83 percent of CPU and takes more than five seconds to respond. Fortunately, recent advances in network compression, approximate computing, and accelerators are enabling deep learning on resource-constrained IoT devices.

The computer chip industry over the last couple of decades has seen its innovation stem from just a few top players like Intel, AMD, NVIDIA, and Qualcomm. In this same time span, the VC industry has shown waning interest in start-up companies that made computer chips. The risk was just too great; how could a start-up compete with a behemoth like Intel which made the CPUs that operated more than 80% of the world's PCs? In areas that that Intel wasn't the dominate force, companies like Qualcomm and NVIDIA were a force for the smartphone and gaming markets. The recent resurgence of the field of artificial intelligence (AI) has upended this status quo. It turns out that AI benefits from specific types of processors that perform operations in parallel, and this fact opens up tremendous opportunities for newcomers.

Facebook abandoned an experiment after two artificially intelligent programs appeared to be chatting to each other in a strange language only they understood. The two chatbots came to create their own changes to English that made it easier for them to work – but which remained mysterious to the humans that supposedly look after them. The bizarre discussions came as Facebook challenged its chatbots to try and negotiate with each other over a trade, attempting to swap hats, balls and books, each of which were given a certain value. But they quickly broke down as the robots appeared to chant at each other in a language that they each understood but which appears mostly incomprehensible to humans. The robots had been instructed to work out how to negotiate between themselves, and improve their bartering as they went along. But they were not told to use comprehensible English, allowing them to create their own "shorthand", according to researchers.

These complex and abstract representations can then be identified anywhere in the image. One drawback to CNN's is that increasing model power requires increased model depth. This increases the number of parameters in the model, lengthening training time and predisposing to the vanishing gradient problem, where gradients disappear and the model stalls in stochastic gradient descent, failing to converge. The introduction of Residual Networks in 2015 (ResNets) solved some of the problems with increasing network depth, as residual connections (seen above in a DenseNet) allow backpropagation to take a gradient from the last layer and follow it through all the way to the first layer. Recognition that CNN's are agnostic to position, but not orientation is important to note.

If you have been following Data Science / Machine Learning, you just can't miss the buzz around Deep Learning and Neural Networks. Organizations are looking for people with Deep Learning skills wherever they can. From running competitions to open sourcing projects and paying big bonuses, people are trying every possible thing to tap into this limited pool of talent. Self driving engineers are being hunted by the big guns in automobile industry, as the industry stands on the brink of biggest disruption it faced in last few decades! If you are excited by the prospects deep learning has to offer, but have not started your journey yet – I am here to enable it.

Catherine Stinson is a postdoctoral scholar at the Rotman Institute of Philosophy, at the University of Western Ontario, and former machine-learning researcher. I wrote my first lines of code in 1992, in a high school computer science class. When the words "Hello world" appeared in acid green on the tiny screen of a boxy Macintosh computer, I was hooked. I remember thinking with exhilaration, "This thing will do exactly what I tell it to do!" and, only half-ironically, "Finally, someone understands me!" For a kid in the throes of puberty, used to being told what to do by adults of dubious authority, it was freeing to interact with something that hung on my every word – and let me be completely in charge. For a lot of coders, the feeling of empowerment you get from knowing exactly how a thing works – and having complete control over it – is what attracts them to the job.

Ongoing developments in neural network models are continually advancing the state-of-the-art in terms of system accuracy. However, the predicted labels should not be regarded as the only core output; also important is a well calibrated estimate of the prediction uncertainty. Such estimates and their calibration is critical in relation to robust handling of out of distribution events not observed in training data. Despite their obvious aforementioned advantage in relation to accuracy, contemporary neural networks can, generally, be regarded as poorly calibrated and as such do not produce reliable output probability estimates. Further, while post-processing calibration solutions can be found in the relevant literature, these tend to be for systems performing classification. In this regard, we herein present a method for acquiring calibrated predictions intervals for neural network regressors by posing the regression task as a multi-class classification problem and applying one of three proposed calibration methods on the classifiers' output. Testing our method on two exemplar tasks - speaker age prediction and signal-to-noise ratio estimation - indicates both the suitability of the classification-based regression models and that post-processing by our proposed empirical calibration or temperature scaling methods yields well calibrated prediction intervals. The code for computing calibrated predicted intervals is publicly available.