The learning direction of FPGA machine learning

After 2 months of knowledge of machine learning. I've found that machine learning has a variety of directions. Page sort. Speech recognition, image recognition, recommender system, etc. Algorithms are also varied. After seeing the other books, I found that except for the K-mean clustering. Bayesian, neural network, online learning and so on, there are a lot of other algorithms. For example: Immune algorithm, genetic algorithm, principal component analysis. Ant colony algorithm and so on.

It seems that very many algorithms are needed to do a lot of research talent with very good. Deep learning is said to have been upgraded by neural networks.

The neural network itself is a book, the content is very much. The Dragon Star program also involves the application of multiple algorithms. is to follow the popular algorithm to learn. Or find the latest machine learning algorithms??

The recent comparison of fire is deep learning. More information. Learn more people. Or is the more uncommon immune algorithm, ant colony algorithm??? From a performance point of view, deep learning performance is very good. But the immune algorithm can develop better in the next 2 years. Under such circumstances, what is better than learning?? I think. Suppose you have advanced mathematical skills, very good thinking. There are a lot of creative friends, and my advice is to develop new algorithms. Like the immune algorithm class. Of course it would be better if we could create a bee-building algorithm. It is expected that very many people do not have this condition, then we will be a trailing person. Choose the more popular deep learning algorithm now. Find a deep learning application for the occasion and company. should also be very good.

I may have something different. I feel like I can do AI, don't want to say that robots defeat humans. There are so many like the robots in the sci-fi movie, I don't have that ability.

I want to do things very easy, let the machine's eyes understand ordinary things, do some simple things can be. So my basic direction is. Machine vision.

So how do I plan to move forward in a step-by-step way? Or do I have to learn something?

My current content is about image processing, the fact that the processing is the most front-end pattern recognition processing work. Let the image of the characteristics of a better embodiment. The next step is pattern recognition, which can only be understood in a narrow sense. is feature extraction. has actually entered the machine learning range. The last is machine learning. To be able to unify cognition. There's a lot of design to be done on an FPGA processor chip (this will be said later). A different angle to explain what I want to learn, image processing, for example, the contrast degree. Image correction. Boundary scan and so on. Machine learning, it is from the numerous learning algorithms inside. Good use of images, for example, deep learning and principal component analysis (some of the other things can be understood.)

Applications, and simple algorithms).

Machine learning can sometimes do the content of image processing. For example, clustering is capable of cutting images. But why do we still have to learn the technology of image processing?? The idea is that machine learning is the process of extracting features on its own initiative, like a decision tree that you might know about its classification process. The process of extracting features. But very often do not know, but image processing is artificial to provide, separate, some special characteristics.

Can reduce the difficulty of machine learning (pure conjecture, and the knowledge of machine learning).

What about the idea of FPGA??? The main consideration is the calculation speed, the current FPGA computing speed is the best, for example: The drone disaster relief, the speed of flight. The pixel of the camera. Identification requires a lot of computing and locating personnel information.

Also, for example, training time, speed is an important indicator.

However, the FPGA complex calculation is not completed. Assume that the GPU or APU will be able to do a higher computational power on that day. I will also consider to study.

These are just basic learning content, as well as a lot of small content to keep up with. Let's say math. The content is very much, I can only in the time I can control, against my ability, to balance each part of the study time. These are the friends that I want to say that want to study with me. Just study with me. My qq,849886241. Ask for attention, ask for help.

The road is very long, need help.

The learning direction of FPGA machine learning