Inspired by a previous attempt to answer crossword questions using neural networks (Hill, Cho, Korhonen, & Bengio, 2015), this dissertation implements extensions to improve the performance of this existing definition model on the task of answering crossword questions. A discussion and evaluation of the original implementation finds that there are some ways in which the recurrent neural model could be extended. Insights from related fields neural language modeling and neural machine translation provide the justification and means required for these extensions. Two extensions are applied to the LSTM encoder, first taking the average of LSTM states across the sequence and secondly using a bidirectional LSTM, both implementations serve to improve model performance on a definitions and crossword test set. In order to improve performance on crossword questions, the training data is increased to include crossword questions and answers, and this serves to improve results on definitions as well as crossword questions. The final experiments are conducted using sub-word unit segmentation, first on the source side and then later preliminary experimentation is conducted to facilitate character-level output. Initially, an exact reproduction of the baseline results proves unsuccessful. Despite this, the extensions improve performance, allowing the definition model to surpass the performance of the recurrent neural network variants of the previous work (Hill, et al., 2015).

Structured scene descriptions of images are useful for the automatic processing and querying of large image databases. We show how the combination of a semantic and a visual statistical model can improve on the task of mapping images to their associated scene description. In this paper we consider scene descriptions which are represented as a set of triples (subject, predicate, object), where each triple consists of a pair of visual objects, which appear in the image, and the relationship between them (e.g. man-riding-elephant, man-wearing-hat). We combine a standard visual model for object detection, based on convolutional neural networks, with a latent variable model for link prediction. We apply multiple state-of-the-art link prediction methods and compare their capability for visual relationship detection. One of the main advantages of link prediction methods is that they can also generalize to triples, which have never been observed in the training data. Our experimental results on the recently published Stanford Visual Relationship dataset, a challenging real world dataset, show that the integration of a semantic model using link prediction methods can significantly improve the results for visual relationship detection. Our combined approach achieves superior performance compared to the state-of-the-art method from the Stanford computer vision group.

I have been in the space of artificial intelligence for a while and am aware that multiple classifications, distinctions, landscapes, and infographics exist to represent and track the different ways to think about AI. However, I am not a big fan of those categorization exercises, mainly because I tend to think that the effort of classifying dynamic data points into predetermined fixed boxes is often not worth the benefits of having such a "clear" framework (this is a generalization of course as sometimes they are extremely useful). I also believe this landscape is useful for people new to the space to grasp at-a-glance the complexity and depth of this topic, as well as for those more experienced to have a reference point and to create new conversations around specific technologies. What follows is then an effort to draw an architecture to access knowledge on AI and follow emergent dynamics, a gateway of pre-existing knowledge on the topic that will allow you to scout around for additional information and eventually create new knowledge on AI. I call it the AI Knowledge Map (AIKM).

The fact that increasingly complex automation is applied to problems that were solved entirely by humans opens the door to great opportunities but also to fallouts and shortcomings. According to consulting firm Accenture, American companies are expected to invest 35 trillion dollars in cognitive technologies before 2035, and that does not take into account other big players, such us Europe, China or Japan. Governments, such the French, are recognizing the importance of AI for the economy and the society. In Spain, the Ministry of Digital Agenda (Minetad) has created a group of experts that is working on a White Paper on AI.

Sony's chief executive says the company is not planning to allow Fortnite players on PlayStation to compete cross-platform because it feels their console offers the best experience for gamers. The technology giant caused controversy earlier this year when the battle royale game launched on the Nintendo Switch, but existing PlayStation players discovered they could not move or change platforms on the same account – despite Xbox and PC players being able to do so. In spite of vocal protests from gamers in response, Sony has not yet changed its policy. New chief executive Kenichiro Yoshida said the company believes the best experience is on PlayStation 4. The technology giant caused controversy earlier this year when the battle royale game launched on the Nintendo Switch, but existing PlayStation players discovered they could not move change platforms on the same account – despite Xbox and PC players being able to do so. Speaking at the IFA technology show in Berlin, Sony chief executive Kenichiro Yoshida said he felt playing on the PlayStation 4 was the best experience for gamers and therefore should not be compromised.

Sharp's newest fridge freezer doesn't have a water fountain or a voice assistant. Instead it houses a vacuum-packing slot that will help keep produce fresh for longer, reduce food waste (leftovers!) and, yes, even prep food for that millennial cooking style of choice -- sous vide. The VacPac Pro fridge-freezer had its debut at IFA 2018, and doesn't require proprietary bags. You can use any sealer bag, and the slot will suck out the air at the touch of a button -- which we proceeded to do on some plastic fruit. The company believes the method can extend the longevity of meat and dairy by up to eight times, and a spokesperson confirmed that the sealed bags are ideal for sous vide-style preparation, ensuring perfect edge-to-edge cooking, retaining the juices, nutrients and other good things.

Deep learning has changed the way visual problems are addressed in machine learning. Elements such as convolutional neural networks (CNN) have now become the standard architecture for areas like image recognition and computer vision. Research in these areas has unveiled scores of new theoretical concepts and innovative practical implementations. A plethora of concepts are available to achieve these futuristic technologies. In this article, we discuss saliency maps, which is one of the most talked-about image recognition concepts now being used in deep learning.

SOFTWARE that can learn is changing the world, but it needs supervision. Humans provide such oversight in two ways. The first is to show machine-learning algorithms large sets of data that describe the task at hand. Labelled pictures of cats and dogs, for instance, allow an algorithm to learn to discriminate between the two. The other form of supervision is to set a specific goal within a highly structured environment, such as achieving a high score in a video game, and then let the algorithm try out lots of possibilities until it finds one that achieves the objective.

I was recently consulting for an organisation that was looking to implement a framework to govern the implementation of Artificial Intelligence (AI) technologies. Like many organisations in their sector, they had been running various'lab' experiments for some time, and had seen positive results; but there was still something holding them back from wholesale investment. A major consulting firm had encouraged them to'accelerate' their innovation by using a framework to govern the roll-out. I asked them where they felt it needed more focus, and they responded saying that it felt somewhat vanilla, a re-hashing of any-old IT project management best practice. "Surely there is something different about AI", they asked?

The Japanese space agency will land two robots on an asteroid next month – the latest step in historic plans to explore its surface and bring samples back to Earth. The mission to the 1km-wide space rock, known as Ryugu, could provide clues not only to the asteroid's formation but to the formation of our solar system. The Japanese space agency have now selected dates for the deployment of smaller crafts from Hayabusa-2 . From the International Space Station, Expedition 42 Flight Engineer Terry W. Virts took this photograph of the Gulf of Mexico and U.S. Gulf Coast at sunset This image of an area on the surface of Mars, approximately 1.5 by 3 kilometers in size, shows frosted gullies on a south-facing slope within a crater. The image was taken by Nasa's HiRISE camera, which is mounted on its Mars Reconaissance Orbiter The Soyuz TMA-15M rocket launches from the Baikonur Cosmodrome in Kazakhstan on Monday, Nov. 24, 2014, carrying three new astronauts to the International Space Station.