Last Friday, we mentioned how Google's artificial intelligence software DeepMind has the ability to teach itself many things. It can teach itself how to walk, jump and run. Or defeat the world's best player of the Chinese strategy game, Go. The science of teaching computers how to do things is called Deep Learning. Offered through Udacity, the course is taught by Vincent Vanhoucke, the technical lead in Google's Brain team.

We live in a world of chatbots. Chatbots improve human interaction with systems by giving a response based on the user input. This means chatbots are simple automated programs that can process simple user inputs and provide a meaningful output. But with time, chatbots evolved and now are impacting the industries around us. We live in a world of chatbots.

It can be hard to predict which films will triumph at the box office and which will completely bomb. Just ask the people who made the recently released sci-fi thriller Valerian, which cost a shade under $200 million to make--and after one weekend, has made back just $23.5 million in total ticket sales. Perhaps in the future, forecasting audience enjoyment won't be such an inexact science. Disney and the California Institute of Technology have teamed up to build an artificial intelligence system that they say can measure moviegoers' facial reactions to determine just how much they're really liking a film. According to a paper published collaboratively by the two institutions last week and first spotted by Digital Trends, the team of researchers had test audiences watch Disney movies and used an infrared high-definition camera to capture their reactions.

All these terms fly around in IT organizations today as CIOs, battling marketplace uncertainties and cost pressures, look for ways to enhance enterprise performance. As with most technology trends, the hype tends to overhang reality by a significant margin in the early stages of adoption, much in line with Gartner's hype cycle theory. Early this year, I wrote a piece that discussed how emerging technologies such as artificial intelligence (AI) and blockchain will drive precision medicine this year. Halfway into the year, the signs are that the use of AI technologies has definitely picked up momentum. A recent study by consulting firm Accenture provides us some interesting data points.

FOR a ten-year-old, Amartya is a thoughtful chap. One Monday morning at the Khan Lab School (KLS) in Mountain View, California, he explains that his maths is "pretty strong" but he needs to work on his writing. Not to worry, though; Amartya has a plan. He will practise grammar online, book a slot with an English teacher and consult his mentor. Later he will e-mail your correspondent to ask for help, too. This is the sort of pluck KLS produces. Its pupils do not have homework or report cards or spend all day in classrooms.

Here is the tech news you need to know this Friday. The firm's rapidly rising expenses brought its operating income down 51% from the same period last year to $628 million (£480 million). The company refreshed its iPod lineup on Thursday, and cut out the only two models that don't run on the iOS operating system. His personal net worth is now estimated to be around $90 billion (£68 billion). The company managed to beat Wall Street's expectations in terms of revenue, however, with $574 million (£438 million) versus an expected $537 million (£410 million).

We study the active learning problem of top-$k$ ranking from multi-wise comparisons under the popular multinomial logit model. Our goal is to identify the top-$k$ items with high probability by adaptively querying sets for comparisons and observing the noisy output of the most preferred item from each comparison. To achieve this goal, we design a new active ranking algorithm without using any information about the underlying items' preference scores. We also establish a matching lower bound on the sample complexity even when the set of preference scores is given to the algorithm. These two results together show that the proposed algorithm is nearly instance optimal (similar to instance optimal [FLN03], but up to polylog factors). Our work extends the existing literature on rank aggregation in three directions. First, instead of studying a static problem with fixed data, we investigate the top-$k$ ranking problem in an active learning setting. Second, we show our algorithm is nearly instance optimal, which is a much stronger theoretical guarantee. Finally, we extend the pairwise comparison to the multi-wise comparison, which has not been fully explored in ranking literature.

PCA is a classical statistical technique whose simplicity and maturity has seen it find widespread use as an anomaly detection technique. However, it is limited in this regard by being sensitive to gross perturbations of the input, and by seeking a linear subspace that captures normal behaviour. The first issue has been dealt with by robust PCA, a variant of PCA that explicitly allows for some data points to be arbitrarily corrupted, however, this does not resolve the second issue, and indeed introduces the new issue that one can no longer inductively find anomalies on a test set. This paper addresses both issues in a single model, the robust autoencoder. This method learns a nonlinear subspace that captures the majority of data points, while allowing for some data to have arbitrary corruption. The model is simple to train and leverages recent advances in the optimisation of deep neural networks. Experiments on a range of real-world datasets highlight the model's effectiveness.

In this work we present a novel system for PET estimation using CT scans. We explore the use of fully convolutional networks (FCN) and conditional generative adversarial networks (GAN) to export PET data from CT data. Our dataset includes 25 pairs of PET and CT scans where 17 were used for training and 8 for testing. The system was tested for detection of malignant tumors in the liver region. Initial results look promising showing high detection performance with a TPR of 92.3% and FPR of 0.25 per case. Future work entails expansion of the current system to the entire body using a much larger dataset. Such a system can be used for tumor detection and drug treatment evaluation in a CT-only environment instead of the expansive and radioactive PET-CT scan.

Single image super-resolution is the task of inferring a high-resolution image from a single low-resolution input. Traditionally, the performance of algorithms for this task is measured using pixel-wise reconstruction measures such as peak signal-to-noise ratio (PSNR) which have been shown to correlate poorly with the human perception of image quality. As a result, algorithms minimizing these metrics tend to produce over-smoothed images that lack high-frequency textures and do not look natural despite yielding high PSNR values. We propose a novel application of automated texture synthesis in combination with a perceptual loss focusing on creating realistic textures rather than optimizing for a pixel-accurate reproduction of ground truth images during training. By using feed-forward fully convolutional neural networks in an adversarial training setting, we achieve a significant boost in image quality at high magnification ratios. Extensive experiments on a number of datasets show the effectiveness of our approach, yielding state-of-the-art results in both quantitative and qualitative benchmarks.