As a child, I was really encouraged by the explorers, and I grew up in INDIA. I wanted to be an explorer when I grew up. As a computer science engineer, I can always look for new things that we can do just that - where possible. Machine learning and research is experimentation, it sounds like psychological testing.
We have seen a tremendous rise in the last five years in what machines can do, by comparison, say, a decade or two years With the advent of more data and more computing power, we can imagine how big and what kind of change in the game we can imagine. rigid, rigid rules are no real solution to the world's problems. So machine learning is about learning from examples.
auto=compress&cs=tinysrgb&dpr=2&w=500 2x" />
Instead of writing 500,000 lines of code, instead the machine learns from the observations of the earth. We look at a bunch of these examples in a machine learning algorithm, maybe millions, maybe billions, maybe even billions, to identify patterns and do the usual from there. In image recognition, we were able to train models. you see a cake and you see a baby, maybe a birthday party. If you see a cake with many children, it is very likely a birthday party. That actually teaches the machine to make ideas that we humans are natural and good at, you see how amazing people are, and how amazing your four-year-old, who can see faces. Machine learning has been the beginning of a major shift in the field of speech recognition.
To teach speech recognition, try to interact with the noisy room, use real-world sounds and mix it with examples we already have, “It's cold outside, it's cold outside?” Now, whatever sound in nature, our speech recognition systems can understand what you are saying. They can distinguish one speaker from another. With machine learning, we now have an algorithm that learns how to imitate a human linguist. Most of the language we see today, is unorganized, "Blah blah blah blah blah blah, blah blah, and they say" OK. "
Emoji and stickers included. Now, with Google, we have reached a point where you can have a very natural conversation. The helper products we create at Google use our best methods for machine learning, image comprehension, natural language comprehension. That is a promising guide to building systems that can be used to navigate the real world. We wanted to make this open source project, so that everyone outside of Google could use the same program we use within Google. 
There are many people who have used it very much, creating a very simple one without a single knowledge and machine learning. So they have ideas. They don't need to do the hard work we have already done. I saw a good example where someone used to have a cat around their house all the time, so they trained the model to pinpoint whenever the cat was present and it would open the sprayers to scare the cat away.
This is an elderly couple in Japan who owned a cucumber farm and one of the biggest jobs is sorting cucumbers, like that, smooth, very thin, straight, curved. It's actually hard work. So the wife spent many hours a day sorting cucumbers, so the son took a computer-based model and was able to create a system of sorting cucumbers and sorting them automatically. All the time spent filtering cucumbers will be used in the best ways. 
387 million people with diabetes are at risk of developing recurrent diabetes. It causes blindness, how you can get the symptoms of diabetes by taking photographs immediately, but they are simply not enough for doctors and it takes hours to interpret. So we trained an algorithm that could read images right there. The algorithm can help doctors get more people tested for the disease.
The more you learn about machine learning and the types of things you can do, the more you see the opportunity to improve people's lives. You can use machine learning to save energy on an important scale, or track the spread of disease and epidemics.
We can use a computer vision model for anyone with poor eyesight. We can make a speech impediment for everyone in the world and greatly improve the knowledge of millions and billions of people.
I don’t see any area of science or human effort that educated systems can help with. If you had asked me a few years ago if a computer would be able to do this anytime soon, I would have said, "I don't think so."It is very empowering to think about what will happen. We think thoughts and do things, you know, no one has ever done them, and it's kind of putting and setting foot in a new mental place here. unresolved problems that will really help people.
Important reading
From helping farmers in Japan filter cucumbers to helping doctors in India as they diagnose eye disease, machine learning is changing the way people use code to solve problems and improve lives. In this video, we will explore how machine learning helps solve many industrial, sector, and resource problems.
0 Comments