ColdFusion - AI Codes its Own ‘AI Child’ - Artificial Intelligence breakthrough!

Starting point is 00:00:00 You are watching Cold Fusion TV. Hi, welcome to another Cold Fusion video. If you've been a follow of this channel for a while, it would be obvious to you that artificial intelligence has made some significant strides in the past couple of years. But today we're going to look at something a little bit different. Google has just created an AI that is capable of creating its own AI that performs better than anything else made before in its field. This sounds impossible, but it's just happened. In this video, we'll take a look. The AI in question is called automatic machine learning or auto ML for short.

Starting point is 00:00:42 It's the work of a team of researchers at Google Brain. The researchers at Google used reinforcement learning to create AI inception as described by Google's CEO Sundar Pichai. Designing better machine learning models, but today it is really time-consuming. It's a painstaking effort of a few engineers and scientists, mainly machine learning PhDs. We wanted to be possible for hundreds of thousands of developers to use machine learning. So what better way to do this than getting neural nets to design better neural nets. We call this approach auto-ML.

Starting point is 00:01:22 It's learning to learn. We are already approaching state of the art in standard tasks like CIFAR image recognition. So whenever I spend time with the team and think about neural nets building their own neural nets, It reminds me of one of my favorite movies, Inception, and I tell them we must go deeper. So instead of spending painstaking hours coding to create an artificial intelligence to do a certain task, AutoML actually automates this process. Here's a quote from Google's research blog. Quote, in our approach, which we call AutoML, a controller neural net can propose a child model architecture

Starting point is 00:02:02 which can then be trained and evaluated for quality on a particular task. That feedback is then used to inform the controller how to improve its proposals for the next round. We will repeat this process thousands of times, generating new architectures, testing them, and giving that feedback to the controller to learn from. In a strange way, you can think of it like a parent giving birth to a child and then teaching that child how to perform a particular task through giving it tests and telling it when it passed or failed. The parent in this case is AutoML and the child is another neural network called N. NASNET. NASNET is an AI capable of recognizing objects such as people, cars, signs, and most things in real-time video.

Starting point is 00:02:45 Now you might be saying that doesn't sound that great, but the most incredible thing here is that this automated AI outperforms any other computer vision that was made before it. NASNAT can perform with 82.7% accuracy, 1.2% better than all other efforts, while being 4% more efficient. An AI built a child AI that performed better than any other artificial intelligence created by humans, and it was more efficient. Just let that sink in. Obvious applications for such a system include self-driving vision support or enabling object

Starting point is 00:03:21 recognition for robots that work in close quarters with humans, such as healthcare for example. This all has huge implications, mainly because if an AI can create systems that are accurate while remaining efficient, it can really take the hard work out of the field of machine learning. AutoML may be one of the first examples, but it does set a powerful precedent. AutoML also makes AI accessible to a wider set of people. As Google puts it, quote, we hope that the larger machine learning community will be able to build on these models to address multitudes of computer vision problems we have not yet imagined. We think that this can

Starting point is 00:03:58 inspire new types of neural nets and make it possible for non-experts to create neural nets tailored to their particular needs, allowing machine learning to have a greater impact to everyone. End quote. There are some questions that naturally need to be asked with auto machine learning technology, that is, if it becomes widely used beyond just computer vision. Some of the concerns are as follows. What if a parent AI passes down a mistake or bias to a child? And furthermore, what if the pace of learning of the child AI develops so quickly that us as

Starting point is 00:04:30 humans have trouble keeping up and then we can no longer understand what it's even doing. These might seem like odd questions today, but it's possible that one day these will be concerns of the future. On a slightly different but related note, InVyre recently released a video of its own AI learning to alter scenes by changing the weather without human input. The results are pretty spectacular. It's really intriguing to think that this is only the infancy of AI. So anyway, that was just a quick video outlining some stuff with computer vision going on at the moment. I just thought it was interesting and thought you might want to know about it. So let me know what you guys think in the comment

Starting point is 00:05:14 section below. If you've just stumbled across this channel, feel free to subscribe. This has been to GoGo, you've been watching Cold Fusion and I'll see you again soon for the next video. Cheers guys, have a good one.

ColdFusion - AI Codes its Own ‘AI Child’ - Artificial Intelligence breakthrough!

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.