 |
| Facebook is using artificial intelligence to understand the videos |
Facebook announced a new plan in hopes of gaining an advantage by training AI on public videos of the platform users.
Teaching AI systems to understand what is happening in videos like humans is one of the biggest machine learning challenges and the biggest potential hack.
Access to training data is one of the biggest competitive advantages of AI. By gathering this resource from millions of users, the technology giants have made advances in various fields.
Although Facebook has trained image processing models from billions of photos collected on Instagram, the company has yet to announce an equally ambitious video comprehension project.
Facebook said that by learning from public videos that cover nearly every country and hundreds of languages, the AI system can not only improve accuracy, but also adapt to our fast-growing world and learn about the nuances of different cultures and regions. And visual cues.
The project is called "learning from video" and it's part of Facebook's larger effort to build machines that learn like humans.
The resulting machine learning models have been used to create new content recommendation systems and review tools, but it will likely do more in the future.
Artificial intelligence that can understand video content can give Facebook unprecedented insight into users 'lives that can analyze users' hobbies, interests, business preferences, clothing, and a myriad of other personal details.
Facebook can access this type of information through its current ad targeting process. However, the ability to analyze videos using artificial intelligence expands its functionality to a variety of data sources.
Although the project is still in an early stage, it is paying off. Facebook said: It is using this technology to improve Instagram Reels recommendations, for example: videos showing people dancing to music.
The system can also improve the results of speech recognition errors, which improves automatic translation and makes it easier to detect hate speech in the video.
Facebook has stated that privacy must be considered when learning videos and wrote in a blog post: We maintain a strong privacy foundation that uses automated solutions to implement high-level privacy.
By integrating this work at the infrastructure level, she added, we can systematically implement data protection requirements throughout the system and support work such as AI, including implementing technical safeguards throughout the data life cycle.
Understanding what is going on in the videos can be a very challenging task for AI systems as there are many obstacles such as background noise that make speech difficult to understand.
However, within a year of starting the video learning project, Facebook takes what the system has learned and puts it into practice in other areas.