In the world of artificial intelligence, there are still many shortcomings to be solved. Most AI solutions do not handle multiple types of input. Specifically, most artificial intelligence tools can only focus on one direction, such as sound, vision, or text. At present, few people try to combine the three aspects to build an AI solution. Why? Because at the current state of the art, we are still far from creating a complete AI system.
But now, someone is finally trying, and a new project from Google and MIT is taking the first step toward a versatile complete AI solution. More specifically, the two companies will work together to develop an AI solution that can simultaneously process sound, text, and images. It is conceivable that once the research project has made a breakthrough, artificial intelligence will be given to human-like multiple senses, which will be a very ambitious research project.
For us humans, it is almost impossible for us to use only one sense at any given time. On the other hand, artificial intelligence does not have this capability at all. Matching what you see and hear is the second talent of human beings, but it is very difficult for the machine to achieve a "sensory".
So, just in terms of senses, having AI with some of the human functions is actually a task that seems impossible, not to mention the level of intelligence of the machine.
In any way, it is not easy to create an algorithm that can learn and adapt like humans. The new research report released by MIT and Google points the way to this attempt, allowing people to see the possibility of giving the AI ​​system multiple “sensoryâ€. The new paper outlines how AI regulates what it hears and sees and synchronizes it, much like the way a human brain works.
丨Yusuf Aytar, postdoctoral researcher at the Massachusetts Institute of Technology
Yusuf Aytar, co-author of the paper and postdoctoral at the Massachusetts Institute of Technology, said: "Whether you have heard the engine sound or saw the car, it doesn't matter, because you can immediately recognize that this is the same concept. This information is already in your The brain is unified."
The key word here is coordination and unification. Instead of teaching the algorithm to new things, the researchers created a way for the algorithm to unify one sense from another. Aytar gave an example to illustrate: When an autonomous car hears the sound of an ambulance, it can be linked to the ambulance. Even if the ambulance is not seen because of the line of sight, it can be avoided in advance.
To train the AI ​​system, MIT researchers first showed some video files with audio to the neural network. When the neural network received the video and audio, it began to try to predict between the object and the sound. contact. The researchers then began entering images with captions in the same algorithm, allowing the neural network to associate objects in the image with their captions. By analogy, this completes the conversion and recognition between video, sound, image and text.
Figure 输入 Input sound, image, and text into the same neural network
Training this system will require a lot of work, although several tests have proven to be quite successful. Now, the algorithm just provides "simple" information, but there is no reason to think that it can't handle more complicated things. The use of this groundbreaking technology will bring new life to the world of human development in the coming years.
Although the tests that have been carried out have been very successful, there is still a lot of work to be done to train this system. Currently, researchers only provide relatively simple information for algorithms, and future training data will become more and more complex. Giving multiple senses to the AI ​​system, this groundbreaking research direction is bound to bring new breakthroughs in the field of artificial intelligence in the next few years. (Bio Valley Bioon.com)
Hydrogel Screen Protector Sheets
Universal Screen Protector, TPU Screen Protector, Hydrogel Protective Film, Mobile Phone Screen Protector, Hydrogel Screen Protector, TPU Protective Film
Shenzhen Jianjiantong Technology Co., Ltd. , https://www.mct-sz.com