Some of mankind’s most complex games, like Go, have failed to challenge AIs from the likes of DeepMind. Reinforcement learning trains algorithms by running scenarios repeatedly with a ‘reward’ given for successes, often a score increase.

Two classic games from the 80s – Montezuma’s Revenge and Pitfall! – have thus far been immune to a traditional reinforcement learning approach. This is because they have little in the way of notable rewards until later in the games.

Applying traditional reinforcement learning typically results in a failure to progress out the first room in Montezuma’s Revenge, while in Pitfall! it fails completely.

One way researchers have attempted to provide the necessary rewards to incentivise the AI is by adding them in for exploration, what’s called ‘intrinsic motivation’. However, this approach has shortcomings.

“We hypothesize that a major weakness of current intrinsic motivation algorithms is detachment,” wrote Uber’s researchers. “Wherein the algorithms forget about promising areas they have visited, meaning they do not return to them to see if they lead to new states.”

Uber’s AI research team in San Francisco developed a new type of reinforcement learning to overcome the challenge.

The researchers call their approach ‘Go-Explore’ whereby the AI will return to a previous task or area to assess whether it yields a better result. Supplementing with human knowledge to guide it towards notable areas sped up its progress dramatically.

If nothing else, the research provides some comfort us feeble humans are not yet fully redundant and the best results will be attained by working hand-in-binary with our virtual overlords.

 AI & >.

The post Uber’s AI beats troublesome games with new type of reinforcement learning appeared first on AI News.

FastStart learns how gamers play and what files are needed to be downloaded first. Microsoft claims this can half the time it takes for a user to get into their new content — if it previously took 30 minutes to download and play, it will now take just 15 minutes.

Xbox has long had a similar ‘Ready to Start’ system where games can be played while the rest downloads in the background. This system required developers to manually configure it during the game’s development, but FastStart is automatic as a platform-level feature.

Furthermore, as you’d expect from a machine learning feature, it’s always learning how to improve.

Jason Ronald, Principal Group Program Manager for Xbox Platform, says: “Since FastStart takes advantage of machine learning, we will continue to improve our algorithm over time getting players into the fun as soon as possible.”

The news is exciting for gamers who are eager to get into their new releases but face increasingly large downloads as a result of 4K textures and bigger worlds being created.

However, FastStart serves as an example of what machine learning is enabling that could be used outside the gaming world. Even shaving a few seconds off smaller downloads helps to improve the experience for consumers.

FastStart will be launching for select games in Xbox GamePass following a system update this month.

An updated list of FastStart supported titles will be available here.

Are you excited about the use of machine learning to prioritise downloads?

The post E3 2018: Xbox FastStart uses machine learning to get into content faster appeared first on AI News.