Artificial intelligence might prevent video streaming buffering
MIT’s new neural network system can pick different algorithms based on network conditions, something that existing algorithms are unable to do successfully
Video buffering and pixilation still remains the torment for those of us who rely a lot on video streaming services to catch up on the newest movies and the latest TV shows. Whatever the factors may be, the irritating stutters can force people to switch to another piece of content, which has a domino effect on viewership and advertising, down the chain. The Massachusetts Institute of Technology (MIT) could perhaps have the answer. MIT believes that using artificial intelligence (AI) can reduce headaches for viewers and streaming services.
What happens thus far is that streaming services use ABR (Adaptive BitRate) algorithms to determine the resolution of the video playback at that time, depending on the changing network conditions. The video you are watching, which is essentially data, is chopped into smaller bits and transmitted sequentially to the device you are watching this on. At the same time, these algorithms also try to create a buffer ahead, so that your streaming experience potentially doesn’t stall even temporarily. Often, this algorithm type fails to determine the changing network conditions quickly enough, which leads to sudden pixilation of the video you are watching. It will also lead to a fairly long pause, if you skip the video too far ahead.
The MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has developed the “Pensieve” neural network, an artificial intelligence (AI) system that uses machine learning to pick different algorithms depending on network conditions. This will allow it to switch between rate-based algorithms or buffer-based algorithms, depending on network conditions. MIT says that the neural network tunes itself over time based on a system of rewards and penalties, which would allow streaming services to customise this for their content—with priorities for buffering or resolution. If the streaming service is able to predict that a user watching a video on YouTube is about to walk into a poor connectivity area, the system will be able to reduce the streaming resolution sufficiently so as to create enough buffer for potentially stutter-free streaming.
“Studies show that users abandon video sessions if the quality is too low, leading to major losses in ad revenue for content providers,” says MIT professor Mohammad Alizadeh. “Sites constantly have to be looking for new ways to innovate.”
Over time, streaming services, including Netflix and Amazon Prime Video, have regularly optimized the streaming algorithms to provide a smoother viewing experience on slower or inconsistent mobile and home broadband connection, but network conditions still play havoc from time to time.
We could eventually see MIT’s “Pensieve” neural network being adopted by streaming platforms such as Netflix and YouTube, and will make streaming higher resolution 4K content easier for streaming platforms.