This week, Apple told us about two new interesting pieces of technology.
- On the one hand, we have ARKit. It makes it easy for developers to build and deploy “augmented reality” applications to the iPhone in your pocket. What is augmented reality? In the visual space, it is a way to add “virtual” objects on top of what you’d normally see. So, for example, a furniture company could make it easy for you to “see” the new furniture in your living room as if you owned it. That’s nice except that looking at your mobile phone is an awkward way to do augmented reality. We’d prefer to have something like “smart glasses”. It is surely coming in the near future.
- On the other hand, Apple released Core ML which is a way for app developers to make use of fancy machine learning techniques without too much effort. Suppose you want to build an app that can look at people’s skin to help diagnose a disease. You could train a model using your PC, and then drag-and-drop this model into your app, and then deploy it to all new iPhones.
Suppose that we had smart watches that measure your pace, your sleep, your heart rate and, one might dream, your glucose level… combine this with Core ML and you might have an app that assesses your health and makes personalized recommendations.
It is interesting to look at the contrast between Apple and the rest of the industry. Google and Microsoft want us to use their services, Apple wants us to buy devices on which we can run apps. This leads to a different focus.
Speaking of Apple, its new iPad Pro uses a processor designed by Apple. It is faster than the Intel processor Apple uses in its cheapest laptops. Given that even the cheapest laptop from Apple is good enough for most people since its cheap laptops are about “mid-range” compared to other vendors, we can say that an iPad Pro could, in theory, replace adequately a laptop for most people, if we only cared about performance. To be fair to Intel, the Apple processor used in its iPad Pro has extra cores and it runs at a much higher clock frequency than the mobile Intel processor used in the laptops (though the laptop has more RAM). Still, we are at the point where top-of-the-line mobile phones and tablets are more powerful than mid-range laptops. It will be interesting to see how the trend goes in coming years.
Brent Smith and Greg Linden review two decades of recommender systems at Amazon.com. Do read it no matter what your scientific level is, it is worth your time. The mathematics can be safely skipped if you want.
A professor was fascinated by the fact that cats that had their spinal cord severed at birth could still walk. After decades, he managed to get some positive results in human beings that had their spinal cord damaged. That is, paralyzed human beings can regain some control of their legs. As a researcher, he was the subject of mockery.
If you combine machine learning and radiology, you can make predictions about one’s health, including longevity.
Digital therapeutics is the idea that you can use software as medicine. I don’t think it is as crazy as it sounds. That you can cure a disease with a pill is not so intuitive, it is only because we are used to it that we take it for granted. I do think we may see software in the future being a strong component of medical therapies. It is much cheaper to deploy software than to distribute and monitor pills.
Sugar is often thought to be bad for you, but a special type of sugar called trehalose seems to protect against cardiovascular diseases when injected in mice. You can find trehalose on Amazon. It is widely used in Japan, so it ought to be safe, at least in modest quantities.