This helps to validate DeepMind's machine learning techniques and the neural network construction behind it. Having proven their mettle in Go, the DeepMind team could now have the confidence (and funding) to tackle more complex AI challenges. ARTIFICIAL INTELLIGENCE (AI) just overcame a new hurdle: learning to play Go, a game considered thousands of times more complex than chess--well enough to beat the greatest human player at his own game. South Korean national Lee Se-dol, one of the world's top Go players, won only one of the five matches against Google's AlphaGo, missing out on the 1-million prize up for grabs in a recent'challenge' held in Seoul. AlphaGo, an AI system developed by Google DeepMind, just bested the best Go-playing human currently alive. This was not supposed to happen.

It may not look like one of Isaac Asimov's robots or sound like HAL from "2001: A Space Odyssey," but artificial intelligence is here, and it is already having a huge impact on how the world works. From the way you shop for a pair of shoes online to how fast a Formula 1 team can push its car's engine, AI is helping businesses across the globe save millions by improving performance and efficiency. Still, problems like trust and security, not to mention fears of the so-called singularity, when artificial intelligence would overtake human thinking, remain hurdles that the technology must overcome before it goes mainstream. AI hit the news this week after a program called AlphaGo, developed by engineers at DeepMind, the AI startup acquired by Google in 2014 for 580 million, defeated the world's No. 1 Go player Lee Sedol. AlphaGo beat Sedol 4 games to 1, claiming a 1 million prize.

Trying to understand what is _really_ going on in the universe.

In science news around the world, the first part of the two-part ExoMars program is on its way to the Red Planet, Google's DeepMind computer program AlphaGo beats the human world Go champion four games to one, China plans to create its own "Defense Advanced Research Projects Agency," the U.S. Environmental Protection Agency announces plans to further limit methane emissions from oil and gas wells, the U.S. Food and Drug Administration green-lights a plan to release mosquitoes in Florida that have been genetically modified to be sterile, and more. Also, German defense minister Ursula von der Leyen, who was accused of plagiarism in her 1990 dissertation, was cleared of misconduct by her degree-granting institution. And a watercolor painting showing the intricate structure of an Ebola virus wins the 2016 Wellcome Image Awards' overall prize.

In this paper we consider a version of the zero-shot learning problem where seen class source and target domain data are provided. The goal during test-time is to accurately predict the class label of an unseen target domain instance based on revealed source domain side information (\eg attributes) for unseen classes. Our method is based on viewing each source or target data as a mixture of seen class proportions and we postulate that the mixture patterns have to be similar if the two instances belong to the same unseen class. This perspective leads us to learning source/target embedding functions that map an arbitrary source/target domain data into a same semantic space where similarity can be readily measured. We develop a max-margin framework to learn these similarity functions and jointly optimize parameters by means of cross validation. Our test results are compelling, leading to significant improvement in terms of accuracy on most benchmark datasets for zero-shot recognition.

We focus on detecting complex events in unconstrained Internet videos. While most existing works rely on the abundance of labeled training data, we consider a more difficult zero-shot setting where no training data is supplied. We first pre-train a number of concept classifiers using data from other sources. Then we evaluate the semantic correlation of each concept w.r.t. the event of interest. After further refinement to take prediction inaccuracy and discriminative power into account, we apply the discovered concept classifiers on all test videos and obtain multiple score vectors. These distinct score vectors are converted into pairwise comparison matrices and the nuclear norm rank aggregation framework is adopted to seek consensus. To address the challenging optimization formulation, we propose an efficient, highly scalable algorithm that is an order of magnitude faster than existing alternatives. Experiments on recent TRECVID datasets verify the superiority of the proposed approach. We focus on detecting complex events in unconstrained Internet videos. While most existing works rely on the abundance of labeled training data, we consider a more difficult zero-shot setting where no training data is supplied.We first pre-train a number of concept classifiers using data from other sources. Then we evaluate the semantic correlation of each concept w.r.t. the event of interest. After further refinement to take prediction inaccuracy and discriminative power into account, we apply the discovered concept classifiers on all test videos and obtain multiple score vectors. These distinct score vectors are converted into pairwise comparison matrices and the nuclear norm rank aggregation framework is adopted to seek consensus. To address the challenging optimization formulation, we propose an efficient, highly scalable algorithm that is an order of magnitude faster than existing alternatives. Experiments on recent TRECVID datasets verify the superiority of the proposed approach

This paper discusses the effect of hubness in zero-shot learning, when ridge regression is used to find a mapping between the example space to the label space. Contrary to the existing approach, which attempts to find a mapping from the example space to the label space, we show that mapping labels into the example space is desirable to suppress the emergence of hubs in the subsequent nearest neighbor search step. Assuming a simple data model, we prove that the proposed approach indeed reduces hubness. This was verified empirically on the tasks of bilingual lexicon extraction and image labeling: hubness was reduced with both of these tasks and the accuracy was improved accordingly.

Automatically recognizing a large number of action categories from videos is of significant importance for video understanding. Most existing works focused on the design of more discriminative feature representation, and have achieved promising results when the positive samples are enough. However, very limited efforts were spent on recognizing a novel action without any positive exemplars, which is often the case in the real settings due to the large amount of action classes and the users' queries dramatic variations. To address this issue, we propose to perform action recognition when no positive exemplars of that class are provided, which is often known as the zero-shot learning. Different from other zero-shot learning approaches, which exploit attributes as the intermediate layer for the knowledge transfer, our main contribution is SIR, which directly leverages the semantic inter-class relationships between the known and unknown actions followed by label transfer learning. The inter-class semantic relationships are automatically measured by continuous word vectors, which learned by the skip-gram model using the large-scale text corpus. Extensive experiments on the UCF101 dataset validate the superiority of our method over fully-supervised approaches using few positive exemplars.

In principle, zero-shot learning makes it possible to train an object recognition model simply by specifying the category's attributes. For example, with classifiers for generic attributes like striped and four-legged, one can construct a classifier for the zebra category by enumerating which properties it possesses --- even without providing zebra training images. In practice, however, the standard zero-shot paradigm suffers because attribute predictions in novel images are hard to get right. We propose a novel random forest approach to train zero-shot models that explicitly accounts for the unreliability of attribute predictions. By leveraging statistics about each attribute’s error tendencies, our method obtains more robust discriminative models for the unseen classes. We further devise extensions to handle the few-shot scenario and unreliable attribute descriptions. On three datasets, we demonstrate the benefit for visual category learning with zero or few training examples, a critical domain for rare categories or categories defined on the fly.

Object recognition systems usually require fully complete manually labeled training data to train the classifier. In this paper, we study the problem of object recognition where the training samples are missing during the classifier learning stage, a task also known as zero-shot learning. We propose a novel zero-shot learning strategy that utilizes the topic model and hierarchical class concept. Our proposed method advanced where cumbersome human annotation stage (i.e. attribute-based classification) is eliminated. We achieve comparable performance with state-of-the-art algorithms in four public datasets: PubFig (67.09%), Cifar-100 (54.85%), Caltech-256 (52.14%), and Animals with Attributes (49.65%) when unseen classes exist in the classification task.