We investigate metric learning in the context of dynamic time warping (DTW), the by far most popular dissimilarity measure used for the comparison and analysis of motion capture data. While metric learning enables a problem-adapted representation of data, the majority of meth- ods has been proposed for vectorial data only. In this contribution, we extend the popular principle offered by the large margin nearest neighbours learner (LMNN) to DTW by treating the resulting component-wise dissimilarity values as features. We demonstrate, that this principle greatly enhances the classification accuracy in several benchmarks. Further, we show that recent auxiliary concepts such as metric regularisation can be transferred from the vectorial case to component-wise DTW in a similar way. We illustrate, that metric regularisation constitutes a crucial prerequisite for the interpretation of the resulting relevance profiles.

High demand for computation resources severely hinders deployment of large-scale Deep Neural Networks (DNN) in resource constrained devices. In this work, we propose a Structured Sparsity Learning (SSL) method to regularize the structures (i.e., filters, channels, filter shapes, and layer depth) of DNNs. SSL can: (1) learn a compact structure from a bigger DNN to reduce computation cost; (2) obtain a hardware-friendly structured sparsity of DNN to efficiently accelerate the DNNs evaluation. Experimental results show that SSL achieves on average 5.1x and 3.1x speedups of convolutional layer computation of AlexNet against CPU and GPU, respectively, with off-the-shelf libraries. These speedups are about twice speedups of non-structured sparsity; (3) regularize the DNN structure to improve classification accuracy. The results show that for CIFAR-10, regularization on layer depth can reduce 20 layers of a Deep Residual Network (ResNet) to 18 layers while improve the accuracy from 91.25% to 92.60%, which is still slightly higher than that of original ResNet with 32 layers. For AlexNet, structure regularization by SSL also reduces the error by around ~1%. Open source code is in https://github.com/wenwei202/caffe/tree/scnn

In this paper, we further study the forward-backward envelope first introduced in [28] and [30] for problems whose objective is the sum of a proper closed convex function and a twice continuously differentiable possibly nonconvex function with Lipschitz continuous gradient. We derive sufficient conditions on the original problem for the corresponding forward-backward envelope to be a level-bounded and Kurdyka-{\L}ojasiewicz function with an exponent of $\frac12$; these results are important for the efficient minimization of the forward-backward envelope by classical optimization algorithms. In addition, we demonstrate how to minimize some difference-of-convex regularized least squares problems by minimizing a suitably constructed forward-backward envelope. Our preliminary numerical results on randomly generated instances of large-scale $\ell_{1-2}$ regularized least squares problems [37] illustrate that an implementation of this approach with a limited-memory BFGS scheme usually outperforms standard first-order methods such as the nonmonotone proximal gradient method in [35].

Bayesian max-margin models have shown superiority in various practical applications, such as text categorization, collaborative prediction, social network link prediction and crowdsourcing, and they conjoin the flexibility of Bayesian modeling and predictive strengths of max-margin learning. However, Monte Carlo sampling for these models still remains challenging, especially for applications that involve large-scale datasets. In this paper, we present the stochastic subgradient Hamiltonian Monte Carlo (HMC) methods, which are easy to implement and computationally efficient. We show the approximate detailed balance property of subgradient HMC which reveals a natural and validated generalization of the ordinary HMC. Furthermore, we investigate the variants that use stochastic subsampling and thermostats for better scalability and mixing. Using stochastic subgradient Markov Chain Monte Carlo (MCMC), we efficiently solve the posterior inference task of various Bayesian max-margin models and extensive experimental results demonstrate the effectiveness of our approach.

Conventional deep neural networks (DNN) for speech acoustic modeling rely on Gaussian mixture models (GMM) and hidden Markov model (HMM) to obtain binary class labels as the targets for DNN training. Subword classes in speech recognition systems correspond to context-dependent tied states or senones. The present work addresses some limitations of GMM-HMM senone alignments for DNN training. We hypothesize that the senone probabilities obtained from a DNN trained with binary labels can provide more accurate targets to learn better acoustic models. However, DNN outputs bear inaccuracies which are exhibited as high dimensional unstructured noise, whereas the informative components are structured and low-dimensional. We exploit principle component analysis (PCA) and sparse coding to characterize the senone subspaces. Enhanced probabilities obtained from low-rank and sparse reconstructions are used as soft-targets for DNN acoustic modeling, that also enables training with untranscribed data. Experiments conducted on AMI corpus shows 4.6% relative reduction in word error rate.

Driven by a declining population, a trend for developing robotic babies has emerged in Japan as a means of encouraging couples to become'parents'. The approaches taken vary widely and are driven by different philosophical approaches that also beg a number of questions, not least whether these robo-tots will achieve the aim of their creators. To understand all of this it is worth exploring the reasons behind the need to promote population growth in Japan. The new Kirobo Mini robot from Toyota (pictured) can't do much but chatter in a high-pitched voice, but supposedly has the smarts of a five-year-old Studies have found that humans and robots can share a'high degree of bonding,' especially with social robots that have a human-like appearance. But, humans may find certain realistic human qualities unappealing.

Nexd today announced seed funding as well as beta availability of its artificial intelligence (AI) and analytics platform, designed to help B2B organizations measure, understand and take action to increase close rates, more effectively forecast sales pipelines and improve overall sales efficiency. The company is addressing one of today's biggest sales management challenges, an action gap, whereby teams frequently miss opportunities to engage and win prospects due to fragmented information spread across thousands of disparate enterprise and SaaS applications and tools. Nexd taps the power of big data, machine learning and predictive analytics to analyze every touch point related to the sales process--including millions of data points from apps and tools outside a company's CRM system--ultimately determining patterns of success as well as where breakdowns occur. The result is deeper intelligence, increased predictability and improved execution across the sales organization. "When we embarked down this path we saw an incredible opportunity to use analytics and artificial intelligence to make people more effective at work," said James Davison, Co-Founder at Nexd. "While we're initially focused on sales, due to its huge impact on the business, we envision the core platform with its guidance and recommendations capabilities will ultimately provide value to many other areas across an organization."

More recently, research has suggested that quantum effects could offer similar advantages for the emerging field of quantum machine learning (a subfield of artificial intelligence), leading to more intelligent machines that learn quickly and efficiently by interacting with their environments. In a new study published in Physical Review Letters, Vedran Dunjko and coauthors have added to this research, showing that quantum effects can likely offer significant benefits to machine learning. "The progress in machine learning critically relies on processing power," Dunjko, a physicist at the University of Innsbruck in Austria, told Phys.org. "Moreover, the type of underlying information processing that many aspects of machine learning rely upon is particularly amenable to quantum enhancements. As quantum technologies emerge, quantum machine learning will play an instrumental role in our society--including deepening our understanding of climate change, assisting in the development of new medicine and therapies, and also in settings relying on learning through interaction, which is vital in automated cars and smart factories."

The advent of ground-breaking genomic technologies in the clinical environment, such as Next Generation DNA Sequencing (NGS), bioinformatics and artificial intelligence, has ushered in a new era in healthcare. Every day, a gigantic amount of genomic information is generated and analysed to identify pathogenic mutations in patient DNA. This new paradigm, known as'Data-Driven Medicine', already allows clinicians to treat the causes of disease rather than its consequences and opens up new treatment opportunities for patients. This has been exemplified in cancer care where therapeutic decisions do not solely rely on morphology, but also on a patient's unique molecular profile. However, the production and analysis of patient genomic profiles is only the first part of the Data-Driven Medicine story.

Computers should be allowed to patent their inventions, experts have claimed. They say without a change in the law, the findings warn that there will be less innovation, caused by uncertainty, which would prevent industry from capitalising on the huge potential of creative computers. They say the future will bring more disputes over inventorship, with individuals taking credit for inventions that are not genuinely theirs, they say. Researchers say uncertainty could stifle innovation unless the changes are made. We are also likely to see disputes over inventorship, they say.