Salakhutdinov's presentation was presented at a machine learning conference called 'NIPS', which has lasted for 31 years and has seen more exponential growth in recent years, attracting nearly 8,000 participants this year.
Most of his presentations centered on the impact machine learning has on autonomous driving car systems, for example, he discussed how to detect cars and pedestrians in busy streets, drive in unknown streets, and project a detailed 3D map of the city.
Salakhutdinov showed Apple's new system that could be used to identify cars and pedestrians and have the car follow the directions on the road for driving, showing how the system works even when it's raining outside, Or when pedestrians and other dangerous situations are out of sight, it can even anticipate where pedestrians might even be able to figure out: "If you asked me five years ago, I would be very skeptical Whether this can be done.
Another project that is on display is the foundation of a research paper Apple published last month that uses laser radar to detect pedestrians and cyclists.
Finally, Salakhutdinov shows how Apple collects data from cars on the road and uses it to create extensive and detailed 3D maps, with information on traffic lights and various road markings, etc. Such maps play an important role in autonomous vehicles effect.
Another project discussed by Salakhutdinov after the show was completed was how to keep the vehicles running the new system moving around the world. He referred to a technology called SLAM that can be used for localization and mapping of real-time locations and maps Synchronization and mapping SLAM is used in robots and autonomous vehicles as well as in mapmaking and augmented reality.
In keeping with Apple's usual love of secrecy, some of this work on developing autonomous systems was surprisingly open, and Tim Cook said the autopilot system was "very exciting," and said that Apple is developing 'all human labor Smart. "Apple also pushed for a reform of the test drive policy for autonomous vehicles, and a recent report said Apple officially tested driverless cars in California.
It is yet to be seen what Apple's efforts in autopilot systems are really about, and Apple's attitude so openly is that it does not really matter how secretive it is, or it releases smoke for the launch of new products. There are also doubts about this.