The Times Bayer bets the farm in 66 billion Monsanto deal: An American pioneer of genetically modified seeds has struck a 66 billion takeover deal with Bayer, the German chemicals giant, in the latest round of megadeal consolidations in agricultural production. Lawyers set to pass M&A paperwork to machines: Luminance, a start-up backed by Mike Lynch, the British technology investor, is aiming to reduce the time taken on due diligence that accompanies multimillion-pound M&A deals from weeks to days or even hours using a new form of artificial intelligence. Public sector job numbers plunge to record low: Employment in the public sector has fallen to a record low as private sector jobs surge ahead despite Brexit, official figures have shown. Mortgage lending falls in wake of Brexit vote: The number of mortgages advanced for house purchases fell in the first month after Britain's decision to leave the European Union. Miner in'bribery scandal' intends to quit AIM: An Africa-focused mining company that has been caught up in a bribery scandal is planning to delist its shares.

The techniques Microsoft Research used to achieve a new world-best error rate will eventually enhance the Cortana Windows 10 personal assistant. Microsoft claims to have achieved the world's lowest error rate for speech recognition, as the company jostles with Amazon, Apple, Google, and IBM to develop products that understand speech as well as humans can. According to Microsoft, its speech scientists at Microsoft Research have achieved a word error rate (WER) of just 6.3 percent under an industry-standard evaluation, using techniques that will eventually enhance Cortana. The previous lowest error rate was 6.9 percent, achieved by IBM's Watson team, which beat their own record of eight percent set last year. Both Microsoft and IBM presented papers detailing their work on speech recognition at the Interspeech conference in San Francisco this week, where papers were also presented by Google's speech researchers.

Is your firm prepared to welcome in an era of legal practice using artificial intelligence (AI)? Where is this technology headed and why must law firm leaders fully understand how cognitive computing works in order to benefit from it? With the vast amounts of structured and unstructured data firms have to deal with, there is increasing pressure from nimble technologies and clients to attract AI and cognitive technologies. Understanding the market implications of digital business information and trends is now key to attaining the best possible performance, revenue gains and cost savings. Berwin Leighton Paisner was one of the first firms to embrace AI in the UK, developing its'contract robot' which promised to complete work in seconds that would have otherwise taken a team of paralegals months to do.

Two themes kept emerging at the inaugural SINET61 cyber innovation conference in Sydney this week. "A strong capability around threat intelligence", said Steve Glynn, chief information security officer (CISO) at ANZ Bank. It'll be about building an understanding of the threat actors, what they're trying to achieve, and what techniques will be needed to defeat them. Threat intelligence was one of the hyper-repeated buzz phrases at SINET61. John Haig, head of security, risk and compliance at Dun & Bradstreet, was one of many speakers who mentioned the other: machine learning.

Gun lovers now have a dating website dedicated to them. Concealed Carry Match allows its users to meet other members who have similar views on Second Amendment rights. The development is potentially controversial, considering the anti-gun rhetoric that has been doing the rounds following the spate of shootings in the United States in the recent past. Working on a similar platform as other niche dating sites, Traverse City, Michigan-based ConcealedCarryMatch.com offers a paid membership, ranging from 10 weekly to 150 for a year. The site launched last month and reportedly registered opening-day traffic of 10,000 interested visitors, signaling at an increased intensification of the debate on gun control ahead of the presidential election in November.

Machine Learning is a great way to use your existing data to identify trends going forward. In this webinar, Gregory Deckler will show us step-by-step how to get started using this very exciting Azure Service, using Power BI and R script. Based on your requests, this webinar will show you how to bring together Azure ML, Power BI, and R in a single solution! Power BI MVP Gregory Deckler will demonstrate how to use R to bridge the gap between Azure ML and Power BI in order to build predictive analytics directly into your data model and reports. Greg is a Director at Fusion Alliance and the Solution Director of Cloud Services.

ELEANOR HALL: Welcome to Techtopia, our segment on the emerging technologies set to disrupt our future; and some of the unexpected questions we may need to ask about them. Joining me in Sydney, as he does every week for Techtopia, is entrepreneur and technology author Steve Sammartino. And also here today in the Sydney studio is AI specialist at the University of Technology Sydney, Professor Mary-Anne Williams. Professor Williams is the director of the Innovation and Enterprise Research Laboratory at UTS, which is also known as The Magic Lab.

Each artificial intelligence (AI) technology is developed to solve a particular problem. Once it solves a problem, however, it's no longer considered AI anymore. Everything from simple calculators and computers, to innovations such as voice recognition on the phone, are just everyday tools that would hardly be considered all that "intelligent". It's almost as if the term AI refers to the science and engineering of making computers do things they can't do yet. Or so suggests Prof Peter Stone from the University of Texas: chair of a panel of academic and industrial thinkers who co-authored the first offering from the 100 Year Study on Artificial Intelligence, an ongoing project hosted by Stanford University to help inform society on how best to manage and respond to developments in smart software, sensors and machines.

Last week, we discussed Multi-class SVM loss; specifically, the hinge loss and squared hinge loss functions. In reality, these values would not be randomly generated -- they would instead be the output of your scoring function f. Let's exponentiate the output of the scoring function, yielding our unnormalized probabilities: Figure 2: Exponentiating the output values from the scoring function gives us our unnormalized probabilities. Figure 4: Taking the negative log of the probability for the correct ground-truth class yields the final loss for the data point. To examine some actual probabilities, let's loop over a few randomly sampled training examples and examine the output probabilities returned by the classifier: Note: I'm randomly sampling from the training data rather than the testing data to demonstrate that there should be a noticeably large gap in between the probabilities for each class label.

Last week, we discussed Multi-class SVM loss; specifically, the hinge loss and squared hinge loss functions. A loss function, in the context of Machine Learning and Deep Learning, allows us to quantify how "good" or "bad" a given classification function (also called a "scoring function") is at correctly classifying data points in our dataset. In fact, if you have done previous work in Deep Learning, you have likely heard of this function before -- do the terms Softmax classifier and cross-entropy loss sound familiar? I'll go as far to say that if you do any work in Deep Learning (especially Convolutional Neural Networks) that you'll run into the term "Softmax": it's the final layer at the end of the network that yields your actual probability scores for each class label. To learn more about Softmax classifiers and the cross-entropy loss function, keep reading.