The subsets of Artificial Intelligence (AI) have multiplied and carry out various tasks that only humans could do. Technologies like Machine Learning carry out administrative tasks, recognize faces, play chess, and even translate languages.
Undoubtedly, the arrival of the AI decade has brought many beneficial developments. Furthermore, Deep Learning learns from unstructured data to compile analytical reports or carry out tasks unsupervised by humans.
All these developments have set the stage for different companies to come into play and prove their worth. As a result, companies like DeepMind were founded to continue developing this field. What is there to know about this company? Here are important things you need to know about Google‘s DeepMind:
Google DeepMind’s history
DeepMind Technologies was established in 2010 in London, but 4 years after that, Google acquired this company. It’s ownership also changes in 2015 because it was then acquired by Alphabet, Inc and since then, it has been a subsidiary of this company. DeepMind was initially founded by Demis Hassabis, Mustafa Suleyman, and Shane Legg, who are all AI enthusiasts and some regard them as pioneers of deep learning.
Since it was established, DeepMind Technologies has opened research centers in the United States, Canada, and France. It started being recognized by many in 2016 after creating AlphaGo which beat Go’s world champion Lee Sedol.
The game was documented and after people saw this, they began giving credit to this company. Above that, they developed another program called AlphaZero that plays chess, shogi and go best.
DeepMind received quite large financial support because individuals like Scott Banister and Elon Musk also chipped in. That was an addition to the capital they derived from venture capital companies, Horizons Ventures, and Founders Fund.
The founders of DeepMind had a solid presentation to these entities and that’s why they received the funding.
General-purpose learning algorithms
DeepMind is very interesting in general-purpose learning algorithms that won’t only improve this field but will help understand the human brain better.
The company has started doing so by developing systems that can play a wide range of different games. According to John Nielson, an assignment helper at essay writing service, that specialize on college papers, one of the founders mentioned that they believe human-level artificial intelligence can be reached when a program can play different games.
Their strategy is backed by scientific studies that prove that games like chess improve strategic thinking capabilities. By machines learning how to play these complex games, they will attain the capability of thinking and acting strategically.
DeepMind’s general-purpose learning algorithms allow the machine to learn through gamification to try and acquire human-like intelligence and behavior.
Even though the company is keenly interested in machine learning to achieve human intelligence, it also has an objective view on the safety of using these technologies.
To avoid a machine apocalypse, DeepMind developed an open-source testbed to determine if an algorithm has a kill switch when there is undesirable behavior. The open-source testbed is called GridWorld and it ensures that AI remains safe and harmless to itself, developers, and other human beings exposed to it.
DeepMind’s deep reinforcement learning
DeepMind just took deep learning to a whole different level with implementing a very different technology system. The system is called deep reinforcement learning which is entirely independent, unlike regular AI systems.
For example, IBM Watson or Deep Blue was developed with a certain purpose and is programmed to function in the desired capacity only.
DeepMind’s deep reinforcement learning isn’t preprogrammed but learns with experience just like any human being does. In essence, it bases deep learning on a convolutional neural network and pairs that with Q-learning, says specialist at best paper writing service. Their systems are then tested on a variety of video games without being programmed instructions on how to play that game.
Everything is done independently by the system and it learns how to play the video game and, after quite a few attempts, plays better than any human being. There are various games that this system has played and mastered more than the best playing human beings.
Deep reinforcement learning removes any human error that could disturb the efficiency of the gameplay. It hasn’t been used in games only but also a variety of different useful systems that have had an impact on the healthcare industry.
WaveNet Collaboration
The WaveNet collaborations have been one of the most remarkable healthcare developments that DeepMind has contributed to. There are millions of people that suffer from speech impairment and can’t get back their original voice.
Text-to-speech systems often produce robotic or unnaturally sounding voices. DeepMind collaborated with Google and speech-impaired individuals like Tim Shaw, who suffers from Amyotrophic Lateral Sclerosis (ALS).
The objective was developing a system that sounds like the natural voice of the patient, which may first seem like mission impossible. According to Kate Ross, an essay writer for you at professional writer service, recreating a voice needs hours of audio recordings of that individual reading a particular script.
Unfortunately, people with speech impairment may not have that luxury because they can’t easily form even one sentence. DeepMind worked on an algorithm that requires only a handful of audio recordings to recreate the voice.
After 6 months, the WaveNet collaboration had already worked on Tim’s voice and presented it to him and his family. The results surprised them because it sounded like Tim’s voice before ALS started affecting his speech abilities. You can see the reaction for yourself on YouTube because the whole experience was filmed and uploaded.
Other contributions to Google
There are a lot of developments that DeepMind has had a hand in and a lot were for Google’s AI department. One of the most popular that the vast majority of the population uses daily is personalized app recommendations. DeepMind’s AI system gathers data on your preferences and then recommends apps similar to the ones that you have downloaded before.
A more complex project that they have taken up is creating algorithms that cool down Google’s servers in their data centers.
DeepMind systems have increased the efficiency of those cooling systems and Google has greater plans in store for this company.As mentioned in research of Jack Holton, an editor at writing services and essay writers uk, very soon, users that have devices that run on Android Pie will have features such as adaptive brightness and battery.
Machine learning will assist with energy conservation on these devices by adapting the brightness to current lighting conditions. Also, it will make the operating systems generally easier to use, improving the user experience.
Creating these systems should have been a little more complex because of the small scale of this project. Machine learning systems of this kind generally require larger computing power to function successfully.
The bottom line
DeepMind has made great strides in the Artificial Intelligence field with many useful innovative systems. The contributions that it has made to Google’s AI department are very valuable and have been used on a global scale.
On the other hand, DeepMind has taken on other collaborations such as WaveNet that add value to the lives of the population. The peculiarity of the AI system they use, deep reinforcement learning has made them the company of choice for Google.