A MailOnline investigation into how much personal information Alexa is recording and storing on its users has revealed the smart assistant eavesdrops on housemates' gossip, private conversations about insurance policies - and even the family dog. Amazon insists Alexa can only be activated when the allocated'wake word' is uttered - being Alexa, Computer or Echo. The tech giant - along with Apple's Siri and, until recently, Google's Assistant - says it saves every single interaction a person has with the device to improve the service - with some'unintentional' snippets also being recorded if it mistakes another noise for a'wake word'. However, evidence seen by MailOnline shows this cannot be the case, or the process is fundamentally flawed, as a host of sounds and conversations were recorded without a clear or legitimate wake word being uttered - some when there was not even a human nearby. A MailOnline investigation into these'secret' archives has revealed an eerie snippets of users' friends, families and children being recorded while they were completely unaware - and without a clear or legitimate wake word being uttered.

Deepfake detection is formulated as a hypothesis testing problem to classify an image as genuine or GAN-generated. A robust statistics view of GANs is considered to bound the error probability for various GAN implementations in terms of their performance. The bounds are further simplified using a Euclidean approximation for the low error regime. Lastly, relationships between error probability and epidemic thresholds for spreading processes in networks are established.

Artificial intelligence is transforming business practices and the labor market across all industries, upending how leaders look at structural issues within their companies. While A.I. is often associated with the tech and finance sector, media has and will continue to incorporate artificial intelligence into every facet of the industry in the coming years. According to McKinsey Analytics, within the media sector, A.I. is expected to add the most value to the marketing and sales side of the industry, adding a projected 300 billion in value to that area alone. The strategy and corporate finance subset of the media sector are expected to gross 99 billion dollars in added value from A.I. in the coming years. Many of the areas within the media industry that are projected to gain the most value from A.I., such as the marketing and corporate finance, will be expanding into traditional A.I., which includes machine learning and statistical techniques, such as regression analysis.

IBM and Thomson Reuters are fusing their AI and data expertise to help financial institutions meet regulatory compliance requirements.

A quick look at the tech gifts and gadgets that will make life easier for Mom and just make her smile. Most moms I know "live in service and sacrifice to their children" 24/7. So it never hurts to go the extra mile to make her feel loved and appreciated. And she'd cry a few happy tears if you clean the bathroom without being asked. But if you – like more than 8 out of every 10 Americans – plan to spend a record amount of money on Mom this Mother's Day, starting with this list is a genius move.

This paper proposes a method for machine learning from unlabeled data in the form of a time-series. The mapping that is learned is shown to extract slowly evolving information that would be useful for control applications, while efficiently filtering out unwanted, higher-frequency noise. The method consists of training a feedforward artificial neural network with backpropagation using two opposing objectives. The first of these is to minimize the squared changes in activations between time steps of each unit in the network. This "temporal smoothing" has the effect of correlating inputs that occur close in time with outputs that are close in the L2-norm. The second objective is to maximize the log determinant of the covariance matrix of activations in each layer of the network. This objective ensures that information from each layer is passed through to the next. This second objective acts as a balance to the first, which on its own would result in a network with all input weights equal to zero.

This past "Star Wars Day," May 4, I watched some of the original trilogy a bit mournfully: Peter Mayhew, who played Chewbacca, passed away the day before. When The Empire Strikes Back took us to the Yoda-dwelling Dagobah, I recalled what the exiled Jedi Master had told premonition-plagued Anakin Skywalker decades earlier, about how to deal with the fear of losing loved ones. "Death is a natural part of life," he tells Anakin. "Rejoice for those around you who transform into the Force. Attachment leads to jealousy--the shadow of greed, that is." Yoda is often held up as an avatar or icon of sagacity.

Apple will be the subject of an anti-competition investigation by the European Union according to a report by the Financial Times. The investigation will focus on allegations by music-streaming platform, Spotify, who filed a complaint with the EU in March. According to the company, Apple -- which offers its own music streaming service called Apple Music -- has unfairly used the popularity of its platform to put Spotify and other companies like it at a disadvantage. This includes it's operating system, iOS, and the App Store. Spotify says Apple puts companies at a disadvantage by leveraging its App Store and iOS to lock companies out and charge them lofty fees.

This week is starting off strong for deals. If you make a purchase by clicking one of our links, we may earn a small share of the revenue. However, our picks and opinions are independent from USA Today's newsroom and any business incentives. There is no better way to ease into the work week than with a little online window shopping. And if you're going to distract yourself from your work or your to-do list, you might as well start with what's on sale, right?

Domain adaptation provides a powerful set of model training techniques given domain-specific training data and supplemental data with unknown relevance. The techniques are useful when users need to develop models with data from varying sources, of varying quality, or from different time ranges. We build CrossTrainer, a system for practical domain adaptation. CrossTrainer utilizes loss reweighting, which provides consistently high model accuracy across a variety of datasets in our empirical analysis. However, loss reweighting is sensitive to the choice of a weight hyperparameter that is expensive to tune. We develop optimizations leveraging unique properties of loss reweighting that allow CrossTrainer to output accurate models while improving training time compared to naive hyperparameter search.