More than half of today's insurance companies use machine learning for predictive analytics, according to a new report by Earnix, an analytics software provider for the financial services industry. Roughly 200 insurers were surveyed as part of Earnix's global "Machine Learning: Growing, Promising, Challenging" study, and they were prompted to select all business areas applicable to them. In total, 70% deployed the technology for risk modeling, the study found. Other uses include the creation of demand models, which forecast market appetite for new products, at 45%. Fraud detection was the third-most popular application, used by 36%.

EMILY BORGHARD has a computer inside her skull, but you wouldn't know it to look at her. A small bump behind her left ear, the only external evidence of her implant, is partially covered by a tuft of hair that's still growing in from the last time she had the batteries changed. Before Borghard received a brain implant, she was having as many as 400 "spikes" of seizure-like activity a day, along with multiple seizures. This unrelenting storm of abnormal neural activity turned her teenage years into a semiconscious nightmare. She couldn't drive a car, attend classes or be left alone for more than half an hour.

Artificial intelligence may still sound like a distant future, but don't be fooled. Social media users are already living AI-fueled lives online, and studies show they're eager for marketers to take the technology even further. A recent Weber Shandwick survey found that many more people view AI's advent as a good thing rather than a bad one -- both for society and for themselves. Forty-five percent of respondents foresee a positive societal impact; just seven percent see a negative one. When it comes to their personal lives, 52% expect AI to benefit them; again, only seven percent said it'll make things worse.

One in five people are concerned their jobs could come into competition with automated or artificial intelligence technologies, according to a survey commissioned by tech provider Ricoh Ireland. The study also revealed that just 29 per cent of people believe schools are equipping children with the necessary skills for the digital era. Only a third felt schools have access to the range of modern technologies to help students become digitally proficient. The survey found that confidence in Irish third-level education was higher, with 64 per cent believing graduates are digitally prepared to enter the workplace. Nevertheless, half of respondents said the Irish education system lags behind its European counterparts.

Waymo is hoping to expand its vehicle automation efforts to delivery trucks in a challenge to similar plans by rival firm Uber. The self-driving car unit of Google parent Alphabet is gambling that autonomous vehicles will take over long distance haulage in the coming years. But their plans would still see human drivers handling local pickup and delivery routes. Waymo is hoping to expand its efforts to delivery trucks in a challenge to similar plans by rival firm Uber. A Goldman Sachs Economics Research report issued earlier this month found that autonomous vehicles will take large numbers of jobs away from drivers in the US.

More than half of insurance firms around the world have implemented machine learning technology into their business. The claim comes from Earnix, a leading provider of analytics solutions for the financial services industry, who announced the results of a global survey of more 193 insurance executives today (May 31). Earnix said the survey shows wide adoption of Machine Learning (ML) across the globe, and the expectation that ML will bring "significant" change to the industry over the next 3-5 years. Results revealed that over half (54%) of the executives surveyed said that their organisation was using Machine Learning for predictive analytical modelling. Of those deploying the technology, 70% said they were using it for risk modelling, followed by demand models (45%) and fraud detection (36%).

Google is using machine learning to teach computers to sketch and make music, but one engineer says it isn't ready to "generate" a new Beatles album just yet. IN the future, cars will drive themselves, fridges will order groceries, and doors will unlock automatically as you approach. But what happens when computers move beyond chores and take on creative endeavours? What happens when computers start making art? It's a question Google is investigating, not only investing money in making computers code the most efficient programs themselves, but asking them to learn how to draw, and make their own music based on our own.

Artificial intelligence technology is accelerating forward at a blistering pace, and a trio of scientists are calling for more accountability and transparency in AI, before it's too late. In their paper, the UK-based researchers say existing rules and regulations don't go far enough in limiting what AI can do – and recommend that robots be held to the same standards as the humans who make them. There are a number of issues, say the researchers, that could lead to problems down the line, including the diverse nature of the systems being developed and a lack of transparency about the inner workings of AI. "Systems can make unfair and discriminatory decisions, replicate or develop biases, and behave in inscrutable and unexpected ways in highly sensitive environments that put human interests and safety at risk," the team reports in their paper. In other words: how do we know we can trust AI? Even before we get to the stage of the robots rising up, AI that's unaccountable and impossible to decipher is going to cause issues – from problems working out the cause of an accident between self-driving cars, to understanding the reasons why a bank's computer has turned you down for a loan.

Over at Simply Stats Jeff Leek posted an article entitled "Don't use deep learning your data isn't that big" that I'll admit, rustled my jimmies a little bit. To be clear, I don't think deep learning is a universal panacea and I mostly agree with his central thesis (more on that later), but I think there are several things going on at once, and I'd like to explore a few of those further in this post. Jeff takes a look at the performance of two approaches to classify handwritten 0s vs. 1s from the well known MNIST data set. He compares the performance of a 5-layer neural net with hyperbolic tangent activations to the Leekasso, which just uses the 10 pixels with the smallest marginal p-values. He shows, perhaps surprisingly, that the Leekasso outperforms the neural net when you only have a dozen or so samples. Don't use deep learning if you have 100 samples because the model will overfit and you will get bad out of sample performance.

Explain what regularization is and why it is useful. What are the benefits and drawbacks of specific methods, such as ridge regression and LASSO? Explain what a local optimum is and why it is important in a specific context, such as k-means clustering. What are specific ways for determining if you have a local optimum problem? What can be done to avoid local optima?