Google buys machine vision startup focusing on'instant object recognition' It's a good time to be a machine learning startup. Two weeks after Twitter bought, has purchased . The acquisition was made for an unknown sum, and seems primarily a grab for talent. Moodstocks' engineers and researchers will move to Google's Paris R&D site, and the startup's primary commercial product -- an image recognition API for smartphones -- will be phased out. "Ever since we started Moodstocks, our Read More ... Tags: Computer systems Artificial intelligence Machine intelligence Machine learning Places: Americas North America United States Google today announced it has acquired French machine learning startup Moodstocks for an undisclosed sum. The deal is expected to close in the next few weeks and seems to be focused primarily on the talent, with the team at Moodstocks moving to Google's Paris R&D site, and its image recognition API for smartphones to be gradually phased out.

July 6, 2016, Over a third (36%) of all developers who are actively working on Big Data or advanced analytics projects now use elements of machine learning according to Evans Data's recently released Big Data and Advanced Analytics survey report. While the market for machine learning is still fragmented, those developers actively working with machine learning are most likely to be targeting financial sectors, Internet of Things, or manufacturing. The survey of over 500 developers actively working with Big Data also showed that decision trees are the most used analytical model which links in closely with artificial intelligence and machine learning development. Linear regression and logistics regression were the next most cited analytical models. Logistics, distribution, or operations were the company departments most likely to be using advanced data analytics or Big Data solutions.

"BioTime harnesses the largest collection of highest-quality gene expression data coming from scrupulously designed and controlled cell differentiation experiments we have seen to date. It was large enough to train a complex architecture of deep neural networks to work as a classifier and a predictor of the embryonic state. We recently tested Embryonic.AI using mouse data and noticed surprising results showing the capabilities of this system in cross-species analysis. Research projects using Embryonic.AI may transform our understanding of cancer and other diseases and possible developments in reinforcement learning may help navigate and control cellular differentiation states", said Alex Zhavoronkov, Ph.D., CEO of Insilico Medicine, Inc. The system utilizes a sophisticated architecture of multi-class deep neural networks (DNNs) and DNN ensembles trained on thousands of samples of carefully selected cells of multiple classes: embryonic stem cells, induced pluripotent stem cells, progenitor stem cells, adult stem cells and adult cells to recognize the class and embryonic state of the sample, achieving high accuracy in simulations.

Linguists and primatologists have joined forces to create the groundwork for'primate linguistics,' helping to decipher the meanings behind monkey speech. The comprehensive study examines the calls of different species, analyzing the structure and placement of these vocalizations, and explains what individual calls and sequences mean. While the language of primates may not be as complex as our own, researchers say these animals demonstrate linguistic capabilities that are both'exciting and sometimes challenging.' Linguists and primatologists have joined forces to create the groundwork for'primate linguistics,' helping to decipher the meanings behind monkey speech. The study, led by an international team of researchers, was published recently in the journals Natural Language & linguistic Theory, and builds on earlier research.

Deep clustering is a recently introduced deep learning architecture that uses discriminatively trained embeddings as the basis for clustering. It was recently applied to spectrogram segmentation, resulting in impressive results on speaker-independent multi-speaker separation. In this paper we extend the baseline system with an end-to-end signal approximation objective that greatly improves performance on a challenging speech separation. We first significantly improve upon the baseline system performance by incorporating better regularization, larger temporal context, and a deeper architecture, culminating in an overall improvement in signal to distortion ratio (SDR) of 10.3 dB compared to the baseline of 6.0 dB for two-speaker separation, as well as a 7.1 dB SDR improvement for three-speaker separation. We then extend the model to incorporate an enhancement layer to refine the signal estimates, and perform end-to-end training through both the clustering and enhancement stages to maximize signal fidelity. We evaluate the results using automatic speech recognition. The new signal approximation objective, combined with end-to-end training, produces unprecedented performance, reducing the word error rate (WER) from 89.1% down to 30.8%. This represents a major advancement towards solving the cocktail party problem.

We consider training probabilistic classifiers in the case of a large number of classes. The number of classes is assumed too large to perform exact normalisation over all classes. To account for this we consider a simple approach that directly approximates the likelihood. We show that this simple approach works well on toy problems and is competitive with recently introduced alternative non-likelihood based approximations. Furthermore, we relate this approach to a simple ranking objective. This leads us to suggest a specific setting for the optimal threshold in the ranking objective.

The goal of any inference algorithm is to recover the hidden parameters related to those k hidden nodes (k may be unknown). Consider a special subset of graphical models, known as latent Gaussian trees, in which the underlying structure is a tree and the joint density of the variables is captured by a Gaussian density. The Gaussian graphical models are widely studied in the literature because of a direct correspondence between conditional independence relations occurring in the model with zeros in the inverse of covariance matrix, known as the concentration matrix. There are several works such as [1,2] that have proposed efficient algorithms to infer the latent Gaussian tree parameters. In fact, Choi et al., proposed a new recursive grouping (RG) algorithm along with its improved version, i.e., Chow-Liu RG (CLRG) algorithm to recover a latent Gaussian tree that is both structural and risk consistent [1], hence it recovers the correct value for the latent parameters. They introduced a tree metric as the negative log of the absolute value of pairwise correlations to perform the algorithm. Also, Shiers et al., in [3], characterized the correlation space of latent Gaussian trees and showed the necessary and sufficient conditions under which the correlation space represents a particular latent Gaussian tree. Note that the RG algorithm can be directly related to correlation space of latent Gaussian trees in a sense that it recursively checks certain constraints on correlations to converge to a latent tree with true parameters.

We present a general framework for classifying partially observed dynamical systems based on the idea of learning in the model space. In contrast to the existing approaches using model point estimates to represent individual data items, we employ posterior distributions over models, thus taking into account in a principled manner the uncertainty due to both the generative (observational and/or dynamic noise) and observation (sampling in time) processes. We evaluate the framework on two testbeds - a biological pathway model and a stochastic double-well system. Crucially, we show that the classifier performance is not impaired when the model class used for inferring posterior distributions is much more simple than the observation-generating model class, provided the reduced complexity inferential model class captures the essential characteristics needed for the given classification task.

We investigate a long-debated question, which is how to create predictive models of recidivism that are sufficiently accurate, transparent, and interpretable to use for decision-making. This question is complicated as these models are used to support different decisions, from sentencing, to determining release on probation, to allocating preventative social services. Each use case might have an objective other than classification accuracy, such as a desired true positive rate (TPR) or false positive rate (FPR). Each (TPR, FPR) pair is a point on the receiver operator characteristic (ROC) curve. We use popular machine learning methods to create models along the full ROC curve on a wide range of recidivism prediction problems. We show that many methods (SVM, Ridge Regression) produce equally accurate models along the full ROC curve. However, methods that designed for interpretability (CART, C5.0) cannot be tuned to produce models that are accurate and/or interpretable. To handle this shortcoming, we use a new method known as SLIM (Supersparse Linear Integer Models) to produce accurate, transparent, and interpretable models along the full ROC curve. These models can be used for decision-making for many different use cases, since they are just as accurate as the most powerful black-box machine learning models, but completely transparent, and highly interpretable.

Based on this thesis statement, Turing devised the "Turing test", which is now considered the standard for qualifying computer programs (chatbots) as intelligent. The test today involves a series of 5 minute-long text conversations with judges, during which the program must convince them that it is human on average at least 30% of the time. This is based on Turing's assumption that by the year 2000, machines would be capable of fooling 30% of human judges after five minutes of questioning. Whether a bot is actually able to pass the Turing test or if Turing did in fact intend the test to be passed is still debatable. Regardless, there are a few bots that have an uncanny humanness to them that will convince a good many people that they are human.