Early recognition of cardiac arrest is vitally important as the chance of survival decreases about 10 percent with each minute. In Denmark AI assistant Corti is listening in to phone calls to emergency services to help detect signs of a heart attack. With Corti implemented, the dispatcher gets a digital assistant that listens in on the conversation and helps to look for important signals in both verbal communication, as well as tone of voice and breathing patterns, while also considering other metadata. All the data provided during the emergency call is automatically analyzed by Corti and then compared to the millions of emergency calls – which Corti has already analysed –to find important patterns. As Corti's understanding of the incident increases, the assistant will try to predict the criticality of the patient's situation based on symptom descriptions and the signals gathered from voice and audio.

Brits can now ask their iThings to give them a brief update on what's happening in the world with the command: "Hey Siri, give me the news." Siri doesn't actually read the news, though, and instead will automatically play the latest podcast from a trusted source of your choice. I was treated to a 2-minute bulletin from BBC News when I said the magic words to Siri this morning, which also offered Sky News and LBC up as alternative sources. As Apple prepares Siri for life inside its HomePod smart speaker, it first added the news briefing feature to the beta version of iOS 11.2.5 -- limiting it to the States at that point, too, where The Washington Post, Fox News, NPR or CNN provide the updates. In a matter of weeks, however, it's now graduated out of beta to become a standard Siri feature in iOS 11, whilst rolling out to new territories.

Global consulting firm Accenture notes that stand-alone, AI-powered digital voice assistant devices are being used "for a range of consumer services such as playing music, turning the heat and lights on and off, and providing news, weather and sports scores." That's to be expected, of course - that's what they're for - but in an online survey of 21,000 consumers in 19 countries, Accenture discovered that digital virtual assistants are becoming "the central hub for home activities in Australia." Again, that's the whole point of digital assistants - they're meant to be the hub of your digital home, not just answering questions but helping you control your other connected devices, but naturally, it's always good to see this being confirmed by actual users. In addition, "three quarters (75%) of these owners said they use their smartphones less for entertainment, more than two thirds use them less for online purchasing and more than half for general information searches (71% and 55%, respectively)." David Sovie, global MD of Accenture's High Tech business said: "Digital voice assistant devices are challenging smartphones as the central hub for all activities in the home. "These low-cost devices deliver valuable and practical benefits and are relatively easy to use, and their rapidly growing popularity is one of the most striking trends in the high-tech industry." Accenture tells us its survey shows that "ownership levels of digital voice assistant devices in Australia are projected to increase more than five times this year, reaching 23% by the end of 2018 from 4% in 2017.

For years, marketers have talked about brands as having personalities. Now they have the tools to bring those brands to life – virtually at least. Rapid developments in artificial intelligence (AI) are being combined with Academy Award-winning animation skills to create virtual humans that are the closest yet to flesh and blood. And for brands, that offers the opportunity to put a very human-looking face on a corporate body. One of the latest iterations of these virtual humans comes from Auckland-based company, Soul Machines, whose co-founder and CEO, Mark Sagar's ground-breaking work in computer-generated faces on films, King Kong and Avatar, was recognised with consecutive Oscars.

Australian-based TrademarkVision developed a deep learning-based reverse visual search platform that protects your brand by identifying similar trademarks from around the world. Simply upload your image to the platform, and their image recognition technology will compare it against other trademarked logos – making it much easier to identify IP infringements than the previous time-consuming and costly text-based search process. "Our technology not only makes it easy for an entrepreneur with a new design to ensure it is unique, but also enables the largest of companies to monitor for infringement," explains Cameron Mitchell, the Chief Operations Officer of TrademarkVision. The young startup has already integrated their technology with the intellectual property departments in the EU, Australia, Chile and more. Most recently, they launched a visual search for industrial designs.

The biggest underwater eruption of the last century took place 600 miles (1,000km) off the coast of New Zealand, scientists have found. The discovery was made after an airline passenger saw a strange substance spreading across the Pacific Ocean in 2012. At the time, scientists identified the material as pumice, a volcanic substance which floats. However, it has taken six years for researchers to understand the scale of the eruption with the help of remotely-operated deep search robots. The findings have been described as a'scientific goldmine' and could increase our understanding of how magma rises from the earth's crust to the surface More than 80 per cent of the volcanoes on Earth are located on the sea-floor.

This finding, which is the result of Polycom's The Changing World of Work research, shows that this percentage is going to increase exponentially over the next 12-18 months. Following on from the survey, Polycom has released their top five drivers that are set to impact business collaboration in the year ahead. The cloud space has transformed, not just our office and workspaces but the way we work. If we look at the next generation of technology – it's modular, it's adaptive, it's solutions based and it is cloud based. As cloud continues to move into the mainstream, the conversation is no longer about that sub 50 office space, the one-to-three user space is back on the table.

This paper presents OptNet, a network architecture that integrates optimization problems (here, specifically in the form of quadratic programs) as individual layers in larger end-to-end trainable deep networks. These layers encode constraints and complex dependencies between the hidden states that traditional convolutional and fully-connected layers often cannot capture. In this paper, we explore the foundations for such an architecture: we show how techniques from sensitivity analysis, bilevel optimization, and implicit differentiation can be used to exactly differentiate through these layers and with respect to layer parameters; we develop a highly efficient solver for these layers that exploits fast GPU-based batch solves within a primal-dual interior point method, and which provides backpropagation gradients with virtually no additional cost on top of the solve; and we highlight the application of these approaches in several problems. In one notable example, we show that the method is capable of learning to play mini-Sudoku (4x4) given just input and output games, with no a priori information about the rules of the game; this highlights the ability of our architecture to learn hard constraints better than other neural architectures.

We consider the task of weakly supervised one-shot detection. In this task, we attempt to perform a detection task over a set of unseen classes, when training only using weak binary labels that indicate the existence of a class instance in a given example. The model is conditioned on a single exemplar of an unseen class and a target example that may or may not contain an instance of the same class as the exemplar. A similarity map is computed by using a Siamese neural network to map the exemplar and regions of the target example to a latent representation space and then computing cosine similarity scores between representations. An attention mechanism weights different regions in the target example, and enables learning of the one-shot detection task using the weaker labels alone. The model can be applied to detection tasks from different domains, including computer vision object detection. We evaluate our attention Siamese networks on a one-shot detection task from the audio domain, where it detects audio keywords in spoken utterances. Our model considerably outperforms a baseline approach and yields a 42.6% average precision for detection across 10 unseen classes. Moreover, architectural developments from computer vision object detection models such as a region proposal network can be incorporated into the model architecture, and results show that performance is expected to improve by doing so.

Interest for the search term'data science,' as measured by Google, over the last five years. In the tech industry, new skills and roles emerge faster than traditional education can keep up with. A recent example is the field of data science and the associated profession, Data Scientist. The simplest definition of the data science field is the practice of collecting, analyzing, and interpreting data -- aided by technology. Most Computer Science degrees do not yet offer Data Science as a major and, as such, many Data Scientists are self-taught. For this reason, it is possible to become a Data Scientist without a formal degree This article will explore what it's like to be a Data Scientist, the skillset required, and how to acquire these skills using mostly free or cheap online resources.