Tokyo-Mitsubishi UFJ to automate 30% of its duties, equivalent to 9,500 jobs

The Japan Times

Bank of Tokyo-Mitsubishi UFJ will automate 30 percent of its duties, equivalent to 9,500 jobs, over the next seven years, using artificial intelligence and other technologies, Kanetsugu Mike, president and chief executive officer of the Japanese lender, said in a recent interview. "We hope to free our bankers from simple work and invest our human resources in fields with high added-value," Mike said, indicating plans to relocate workers to enhance services such as those for wealthy clients. To decrease workloads, the unit of Mitsubishi UFJ Financial Group Inc. has already introduced a computer software program utilizing robotic process automation technology, which can handle complex data-matching processes. The bank is currently using the software in 20 different areas, including screenings of application documents for housing loans, and is considering its use in 250 additional duties. "Through automation, the amount of time each banker can spend with customers will triple," Mike said.

Amazon adds voice recognition to Alexa

Daily Mail

Amazon has added a new voice recognition feature to its Alexa smart assistant. It allows the smart speaker and apps to recognise users from their voice, and personalise information to them. It comes just days after Google, which already has a similar feature in its Assistant, launched a raft of devices in its bid to topple Amazon in the personal AI war. Amazon's new Echo (£89.99 / $99) has six'shells' users can swap to customise its appearance. The next generation speaker, which is powered by Amazon's Alexa voice assistant, will have a dedicated woofer and a tweeter for the first time, as well as Dolby sound.

Bayesian Estimation of Signal Detection Models, Part 1


We begin by calculating the maximum likelihood estimates of the EVSDT parameters, separately for each participant in the data set. Before doing so, I note that this data processing is only required for manual calculation of the point estimates; the modeling methods described below take the raw data and therefore don't require this annoying step. First, we'll compute for each trial whether the participant's response was a hit, false alarm, correct rejection, or a miss. We'll do this by creating a new variable, type: Then we can simply count the numbers of these four types of trials for each participant, and put the counts on one row per participant. For a single subject, d' can be calculated as the difference of the standardized hit and false alarm rates (Stanislaw and Todorov 1999): Its inverse, \(\Phi {-1}\), converts a proportion (such as a hit rate or false alarm rate) into a z score.

Pornhub Will Use AI To Identify, Narrow Searches For Performers

International Business Times

The popular adult site Pornhub announced on Wednesday it will use an artificial intelligence to catalogue videos by tagging more than 10,000 performers through computer vision. "Over the course of the past month alone, while we tested the model in beta, it was able to scan through 50,000 videos that have had pornstars added or removed on the video tags," Pornhub VP Corey Price said in a press release. Pornhub plans to scan and tag over 5 million videos across its platform, eventually categorizing videos according to themes, traits and sex positions as well as performers. This raises new questions about privacy and consent in the porn industry. Pornhub will now use artificial intelligence to identify porn performers and preferences.

You Look Like A Criminal! Predicting Crime With Algorithms Big Cloud Recruitment


Can you really predict if someone is going to commit a crime? Some authorities are using facial recognition, predictive analytics and machine learning to predict who will commit a crime. Even if you can use an algorithm to deduce the likelihood of an individual's future, to apprehend a suspect before a crime is even committed surely cannot lead to a conviction as no offence will have actually taken place. Yes, this is all very Minority Report. Nevertheless, companies are currently working on these technologies to catch the bad guys before they even strike.

Intelligent Automation in the Workplace: What it Is and Why it Matters Ayehu


Imagine sitting down for dinner in an upscale restaurant, waiting for your food and seeing it delivered not by a server, but by a flying drone. In the office realm, robots have been deployed for decades, delivering maximum efficiency and productivity at a minimal cost. But what, exactly, is intelligent automation and why should it matter to your business? Instead of merely carrying out predefined steps and processed, automation powered by artificial intelligence is capable of evolving and improving on its own over time. As tasks and workflows are performed, the data obtained during the process is then used to automatically improve how they are performed in the future.

Researchers find that Google's AI has a higher IQ than Siri


The early version of Siri wasn't terribly powerful, but it was able to answer relatively basic queries and handle rather simple tasks such as setting reminders. In subsequent updates, Siri became much more powerful, even adding contextual awareness along the way. Equally as important, Siri's ability to parse and understand language improved by leaps and bounds as well. Though Siri's capabilities today are far more advanced than the version that shipped six years ago, it's not as if Apple has the market cornered on AI-powered assistants. On the contrary, Siri today faces stiff competition from a number of tech behemoths, including Amazon, Microsoft, and of course, Google.

Building a Neural Net from Scratch in Go


I'm super pumped that my new book Machine Learning with Go is now available! Writing the book allowed me to get a complete view of the current state of machine learning in Go, and let's just say that I'm pretty excited to see how the community growing! In the book (and for my own edification), I decided that I would build a neural network from scratch in Go. Turns out, this is fairly easy, and I thought it would be great to share my little neural net here. All the code and data shown below is available on GitHub.

Deep Blue, DeepStack & Holly - AI in the past, present & future Trade-Ideas


While in 1996 Garry Kasparov won 4 to 2 against IBM s super-computer Deep Blue, in 1997 Deep Blue won against Garry Kasparov. This marked a milestone in the process of computers being able to go on learning and getting more intelligent. Now, 20 years later, we all should realize that Artificial Intelligence is taking over our lives. In March 2017, for the very first time, Artificial Intelligence won Heads-Up No-Limit Texas Hold'em against 33 poker players from 17 different nations. At the University of Alberta in Canada, Matej Moravcik and his team created an Artificial Intelligence machine they call "DeepStack."

What is deep learning?


This second video in the Deep Learning Fundamentals playlist gives an intro to what deep learning is and how it is used.