AI gives a future version of almost everything, using scenes that are computer-generated.
Predicting future actions is now made possible through machine-learning. With more development to come, technology will be able to introduce autonomous robots in various fields.
DeepMind, the artificial intelligence company, is currently using this technology through 3-Dimension computer imaging.
When the machine is exposed to an image or any data, it detects various features from the provided data. It uses GQN technology to foresee or visualize the future. Everything is done only through observation without human intervention.
Joshua Tenenbaum, a professor at Massachusetts Institute of Technology, says that there is a long way to go, but the entire aspect is exciting and interesting. He adds that robotic perception will grow and become more adaptive and this is a step in the right direction.
Computer vision has been improving with rapid strides. Though much of this is achieved through supervised learning, it suffers from two drawbacks, states Ali Eslami, a researcher at DeepMind. Supervised learning requires human supervision and a large dataset that is expensive.
Machine learning is an entirely new and complex technology. With Generative Query Network, the computer program can build a mental picture of the world without human supervision. It just takes a look at the scene from various angles. It can describe the scenario by itself, from another angle.
This technology will lead to a higher level of artificial intelligence. Machines will be able to describe the world with greater sophistication. GQN will gain flexibility and be able to create models for every object required, says Eslami.
This computer vision will be useful for the future world by providing better security operations and self-driving cars that will make the world a safer place. Even human activities can be recognized through this technology, claims the team of researchers at MIT who are in the Computer Science and Artificial Intelligence Laboratory.