Posted February 26, 2015 Google has invented an artificial intelligence that learns how to beat video games, from scratch, without any help or direction. The experiment, from Google-owned DeepMind Technologies, is designed to show how intelligent AIs can figure out complex tasks for themselves. The company released some info today, including these two GIFs, which show the AI learning how to beat Breakout. This is the AI, when first confronted with the problem:And here it is after playing a few hundred times:According to Google, the AI learned how to beat 49 games on the Atari 2600. The aim is to show how computers can learn even more complex tasks, like driving cars. DeepMind was created by former game developer Demis Hassabis, but he has designs way beyond video games. "If this can drive the car in a racing game, then potentially, with a few real tweaks, it should be able to drive a real car," he said.UPDATE : Without being given any rules or prior information, a simple computer has learnt how to play 49 classic Atari games in just two weeks - and it's learnt to play them pretty damn well. But what's most impressive is that the Google-built algorithm it uses wasn't even built specifically to play games, just to learn from its own experience. What does that mean, other than the fact computers can now beat us at Space Invaders and Breakout, as well as Chess, Texas hold'em poker and solving Rubik's Cubes? It turns out we now have the early stages of a general learning algorithm that could help robots and computers to become experts at any task we throw at them, and that's a pretty huge deal. "This is the first time that anyone has built a single general learning system that can learn directly from experience to master a wide range of challenging tasks," Demis Hassabis, one of the lead researchers, told William Herkewitz from Popular Mechanics. Hassabis was one of the co-founders of DeepMind Technologies, the company that started making the algorithm and was bought out by Google last year for a reported US$400 million. Publishing today in Nature, the team explains how the deep learning algorithm, which is called Deep Q-Network, or DQN, was able to master games such asBoxing, Space Invaders and Stargunner without any background information. This includes details such as what "bad guys" to look out for, and how to use the controls. It only had access to the score and the pixels on the screen in order to work out how to become an expert player. By playing the games over and over and over again, and learning from its mistakes, the algorithm learn first how to play the game properly, and then, within a fortnight, how to win. Of course, this isn't the first program that teaches a computer to become an expert gamer. Just over 20 years ago, a program known as TD-Gammon mastered Backgammon. But the difference is TD-Gammon never managed to do that well with similar games, such as Chess and Checkers, as Toby Walsh, a computer scientist from National ICT Australia and UNSW who wasn't involved in the research,explains over at The Conversation. The DQN algorithm, on the other hand, could master a range of different games, thanks to two technological advances. First of all, DQN relies on a positive-reinforcement learning method called Q-learning. This basically means that the algorithm will do everything it can - press every button and move the joystick around like a crazy person - in order to get closer to "Q", which is a value that computer scientists have set as the ultimate reward. In the case of this experiment, that reward was game score, and the higher the better. As Herkewitz explains for Popular Mechanics, this isn't as easy as it sounds: "To understand how to maximise your score in a game like Space Invaders, you have to recognise a thousand different facts: how the pixilated aliens move, the fact that shooting them gets you points, when to shoot, what shooting does, the fact that you control the tank, and many more assumptions, most of which a human player understands intuitively. And then, if the algorithm changes to a racing game, a side-scroller, or Pac-Man, it must learn an entirely new set of facts." But this is where the second improvement comes in - DQN is built upon a network that was inspired by the human brain's ability to separate background noise from important information. Which means DQN is able to gulp up valuable clumps of information based on its prior experience, and learn from them. While this is an awesome breakthrough, it's important to note that this isn't a true general learning algorithm just yet. Programmers still had to set a Q value for the program in order for it to learn - a truly intelligent system would be able to work out its own objectives in order to master a new skill. And DQN never truly understands the games it's playing, like a human would, it just learns what to do in order to get a better score. Because of this, there were some games that DQN couldn't master, such as Montezuma's Revenge (you can read more about these over at The Washington Post). In the future, the team hope to expend the algorithm so that it can help to sift through large amounts of scientific data, and come to its own conclusions. "This system that we've developed is just a demonstration of the power of the general algorithms," one of the developers, Koray Kavukcuoglu, told Herkewitz. "The idea is for future versions of the system to be able to generalise to any sequential decision-making problem." Find out more about how DQN learns in the Nature video below, and go and test your own Atari skills here. Source : http://www.bloomberg.com/news/articles/2015-02-25/google-s-computers-learn-to-play-video-games-by-themselves Spoiler Code of Conduct : http://linustechtips.com/main/topic/95599-linus-tech-tips-forum-code-of-conduct/ LTT beginners guide : http://linustechtips.com/main/topic/334934-ltt-beginners-guide/ Read This Before Posting http://linustechtips.com/main/topic/11724-read-before-posting-in-this-section/ Spoiler AMD 5000 Series Ryzen 7 5800X| MSI MAG X570 Tomahawk WiFi | G.SKILL Trident Z RGB 32GB (2 * 16GB) DDR4 3200MHz CL16-18-18-38 | Asus GeForce GTX 3080Ti STRIX | SAMSUNG 980 PRO 500GB PCIe NVMe Gen4 SSD M.2 + Samsung 970 EVO Plus 1TB PCIe NVMe M.2 (2280) Gen3 | Cooler Master V850 Gold V2 Modular | Corsair iCUE H115i RGB Pro XT | Cooler Master Box MB511 | ASUS TUF Gaming VG259Q Gaming Monitor 144Hz, 1ms, IPS, G-Sync | Logitech G 304 Lightspeed | Logitech G213 Gaming Keyboard | PCPartPicker Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 this is like the nvidia drive thing nvidia made to learn self driving and accident avoidance NEW PC build: Blank Heaven minimalist white and black PC Old S340 build log "White Heaven" The "LIGHTCANON" flashlight build log Project AntiRoll (prototype) Custom speaker project Spoiler Ryzen 3950X | AMD Vega Frontier Edition | ASUS X570 Pro WS | Corsair Vengeance LPX 64GB | NZXT H500 | Seasonic Prime Fanless TX-700 | Custom loop | Coolermaster SK630 White | Logitech MX Master 2S | Samsung 980 Pro 1TB + 970 Pro 512GB | Samsung 58" 4k TV | Scarlett 2i4 | 2x AT2020 Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Don't put these in MP games against humans... Not even hackers will be able to beat them. Awesome Wallpapers made by yours truly! | TECH THEMED WALLPAPERS! What happened to your profile picture and who is this Tanku Neko guy? Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 I'm not completely sure how to feel about this. In a nut shell... Hurray for AI Updated 2021 Desktop || 3700x || Asus x570 Tuf Gaming || 32gb Predator 3200mhz || 2080s XC Ultra || MSI 1440p144hz || DT990 + HD660 || GoXLR + ifi Zen Can || Avermedia Livestreamer 513 || New Home Dedicated Game Server || Xeon E5 2630Lv3 || 16gb 2333mhz ddr4 ECC || 2tb Sata SSD || 8tb Nas HDD || Radeon 6450 1g display adapter || Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Imagine in 100s of years when these A.I complain about having humans on their team. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Yay one step closer to skynet Thats that. If you need to get in touch chances are you can find someone that knows me that can get in touch. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 The Singularity is upon us. Quickly, head for the bunkers! Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 This is awesome!Is there a recording of all attempts somewhere? It would be interesting to see the progression of problem solving. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Imagine in 100s of years when these A.I complain about having humans on their team. No complaining will occur. The human will simply "slip and fall" in the shower... Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 This has already been done, Google's late to the party: 2nd part: https://www.youtube.com/watch?v=YGJHR9Ovszs 3rd part: https://www.youtube.com/watch?v=Q-WgQcnessA I am conducting some polls regarding your opinion of large technology companies. I would appreciate your response. Microsoft Apple Valve Google Facebook Oculus HTC AMD Intel Nvidia I'm using this data to judge this site's biases so people can post in a more objective way. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 The Economist had an interesting chart about this: http://www.economist.com/news/science-and-technology/21645108-you-can-teach-computer-play-games-better-it-teach-itself-computers Centipede, Ms. Pacman and Asteroids are still best played by humans. And that article also has the link to the whole Nature piece in which the findings were published. (Here if you don't want to click through: http://www.nature.com/nature/journal/v518/n7540/full/nature14236.html) Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Yay one step closer to skynet You know Skynet is real, right? The make military satellites. Robots in space... 12600k | MSI MEG S280 | SSUPD Meshilicious | Asus ROG STRIX Z690-I | Crucial 16GB 4800MHz CL38 | MSI Gaming 980Ti | CM V850 SFX | WD SN850 1TB, WD SN550 1TB Pi 4TB NAS | Asus VG27AQ, Asus PB278Q | Logitech G Pro X Superlight | Glorious G-HXL-STEALTH | Keychron K4 V2 | Sennheiser HD 599 w/ Fiio E10 Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 The judgement day is upon us AI in Sarah Connor Chronicles learnt how to play a game too. It mastered Chest. *takes pickaxe and builds nuclear shelter* Connection: 200mbps / 12mbps 5Ghz wifi My baby: CPU - i7-4790, MB - Z97-A, RAM - Corsair Veng. LP 16gb, GPU - MSI GTX 1060, PSU - CXM 600, Storage - Evo 840 120gb, MX100 256gb, WD Blue 1TB, Cooler - Hyper Evo 212, Case - Corsair Carbide 200R, Monitor - Benq XL2430T 144Hz, Mouse - FinalMouse, Keyboard -K70 RGB, OS - Win 10, Audio - DT990 Pro, Phone - iPhone SE Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 AI in Sarah Connor Chronicles learnt how to play a game too. It mastered Chest. The difference being that The Turk eventually became John Henry, the good AI that helps John Connor fight Skynet Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Don't put these in MP games against humans... Not even hackers will be able to beat them. No but everyone will own it the first few thousand matches then we're screwed. I'd like to see it try to learn MWO it would take it years of being owned and probably still never defeat a all human team. | CPU: i7-4770K @4.6 GHz, | CPU cooler: NZXT Kraken x61 + 2x Noctua NF-A14 Industrial PPC PWM 2000RPM | Motherboard: MSI Z87-GD65 Gaming | RAM: Corsair Vengeance Pro 16GB(2x8GB) 2133MHz, 11-11-11-27(Red) | GPU: 2x MSI R9 290 Gaming Edition | SSD: Samsung 840 Evo 250gb | HDD: Seagate ST1000DX001 SSHD 1TB + 4x Seagate ST4000DX001 SSHD 4TB | PSU: Corsair RM1000 | Case: NZXT Phantom 530 Black | Fans: 1x NZXT FZ 200mm Red LED 3x Aerocool Dead Silence 140mm Red Edition 2x Aerocool Dead Silence 120mm Red Edition | LED lighting: NZXT Hue RGB | Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 teach it to play cs go it will have to learn how to work as a team with other players Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 The Singularity is upon us. Quickly, head for the bunkers! Run while i join them to create the best combination of nature and technology. /s This sounds cool, would be awesome if it would end the good way with cooperation instead of the sci-fi annihalation protocol. I would hook up to it though, just out of curiousity May the light have your back and your ISO low. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 The difference being that The Turk eventually became John Henry, the good AI that helps John Connor fight SkynetDid Henry really went to future for that? I thought the liquid metal robots who came through the time portal saying "The answer is NO" were associated with Henry. I watched it again (2 episodes a day) in September. But i never thought about this The tv show had some great moments and i liked the actors who played John and Sarah, they fit so well imho. Cant say that about T:Genesis which is coming out soon..... Connection: 200mbps / 12mbps 5Ghz wifi My baby: CPU - i7-4790, MB - Z97-A, RAM - Corsair Veng. LP 16gb, GPU - MSI GTX 1060, PSU - CXM 600, Storage - Evo 840 120gb, MX100 256gb, WD Blue 1TB, Cooler - Hyper Evo 212, Case - Corsair Carbide 200R, Monitor - Benq XL2430T 144Hz, Mouse - FinalMouse, Keyboard -K70 RGB, OS - Win 10, Audio - DT990 Pro, Phone - iPhone SE Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 When google can make an AI that can beat me at PVP in MMO's, then maybe i'll worry. Until then, google can get on my level.... Then again, google could make a bot, farm the best gear and gold and become the ultimate wallet warrior. I retract my previous statement, all hail google. My (incomplete) memory overclocking guide: Does memory speed impact gaming performance? Click here to find out! On 1/2/2017 at 9:32 PM, MageTank said: Sometimes, we all need a little inspiration. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Pretty incredible. "The unexamined life is not worth living" - Apology 38a, Socrates Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted February 26, 2015 Any guesses as to how long it takes to become self-aware? My PC specifications are in my profile. Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Posted March 1, 2015 I'm gathering the AI would decide it is part of the PCMR? Link to comment Share on other sites More sharing options... Link to post Share on other sites More sharing options...
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now