Joseph Sirosh is the corporate vice president of the Information Management and Machine Learning (IMML) team in the Cloud and Enterprise group at Microsoft. He is also a maths nut and perhaps a geek. Almost all of what he said was several levels above my pay grade so I will do my best not to disappoint TWire’s erudite readers (and keep it light).
“I did my PHD in 1995 in neural networks. That led to my working on fraud detection [machine learning] software for a major credit card gateway and that led to joining Microsoft where I can pursue my passion of machine learning,” he said.
“Maths applies to data. Software is eating the world (apologies to Marc Andreessen for that saying) but data is drowning the world,” he said.
“30 years ago data was almost all analogue, now it iP addresss attached almost all analogue, now its all digital and it has an pursue my passionm s all digital and much of it has an IP address attached. The intelligence you can derive is amazing,” he said.
The problem with big data, lots of data, Petabytes, Exabytes or the stuff is that it presents an enormous challenge to store and analyse, especially in real time. Microsoft and its Azure Machine Learning and the Azure Gallery is a good place to find out more.
Joseph spoke about the ‘connected dairy cow’ , a project from Fujitsu in Japan. Pedometers were attached to each cow and the bovine walking characteristics analysed via Azure to predict estrus. Not only did the data reveal the best insemination time and improved the pregnancy rate by 67% it helped predict the resultant gender of the offspring. The results were so good that Fujitsu expects this to be rolled out to 500,000 cows soon.
Joseph said that with Azure cloud developers could build immensely complex systems in just a few days and introduced two plucky Adelaide undergraduates who hope to move the highly manual and inaccurate vineyard yield prediction models to the cloud using Azure.
Harry Lucas, 20, Liam Ellul, 23 are part of the Seer Insights team. They described an intelligent system that assists vineyard staff, growers and wineries in improving the accuracy of their yield estimates. The key is to use Azure to find patterns and insights from both the current data and from historical data and shifting that from an analysis of what’s happened in the past to an accurate prediction of what might occur in the future.
Joseph commented that what Seer had done is harness computing power, machine learning and analytics that were not available even a few years ago.
Then we heard from Max Valente from Thought studio a software development company he began in Sydney in 2011 as a think tank developing technology to solve business problems and create business opportunities.
Max developed the first virtual wardrobe where a person stands in front of a screen and tries on virtual clothes. The latest venture (it has two sites working as proof of concept) uses sensors and devices like Kinect to map out a shoppers experience in a store.
“Kinect gives us the gender, approximate age and time someone enter and exits the store. We can marry that with things like the weather, promotional activity and more to build a profile. We then use sensors connected to Raspberry Pi to show where they go in the store, where they linger, what they try on – we collect a lot of data,” Max said. One advantage is immediately apparent – identifying shoplifters by facial and other ‘secret’ Kinect recognition.
“The goal is to enhance a shopper experience, understand their journey, to lay out stores more logically, to know what’s hot and what’s not, and of course to increase sales,” said Max.
“The entire system is only possible due to Microsoft’s Azure cloud, analytics, storage and tools,” Max added.
So finally to Cortana
Cortana is a digital assistant that can be personal (as in running on a PC or smartphone) or used for business. The former relies on getting to know you well and the latter on getting to know the company well – and keeping all secrets.
Cortana analytics is too broad a subject to cover here but Joseph demonstrated a medical support practice that uses Cortana as the natural language interface between nurses/doctors and patients in a call centre environment. The result has been much better care, better analysis, easier follow up and less need to hospitalise.
Joseph said that Cortana could develop perceptual intelligence to benefit any business – helping them to get closer to customers, knowing more about what they do, and what customers want etc.
He finished with a haiku:
The cloud turns hardware into software
The cloud turns software into services
The cloud turns data into intelligence
The cloud turns intelligence into action
I then walked outside into Sydney’s beautiful sunshine – not a [visible] cloud in sight! The cloud turns into rain, rain turns into ...