Google — Neural Networks and ‘Machines That Dream’

What are Neural Networks?

Neural networks form the base of deep learning. It is a sub-field of machine learning where the algorithms are inspired by the structure of the human brain. Neural networks take data and train themselves to recognize the patterns in this data and predict the output for new sets of similar data.

As the name suggests neural networks inspired by the brain but let’s break that down what are the neurons? And in what sense do they link together?

grayscale value one for white pixels
grayscale value zero for black pixels

when I say neuron all I want you to think about is a “Think that holds a number” specifically a number between 0 & 1. It really not more than that, for example, the network starts with a bunch of neurons corresponding to each of these 28 x 28 pixels of the input image which is 784 neuron in total each one of these holds a number that represents the grayscale value of corresponding pixels ranging from zero for black pixels up to 1 for white pixels this numbers inside the neuron called “Activation” and image you have in your mind here is each neuron is light up when it’s activation is the high number so all of this 784 neuron make up the first layer of our network.

Now jumping over to the last layer this has 10 neurons each representing one of the digit activations in these neurons again some numbers between 0 & 1 represent how much the system thinks the given image corresponds to the given digit. There are also a couple of layers in between called the hidden layers. The way the network operates activation of one layer determine the activations in next layer and occurs the heart of the network as an information processing mechanism comes down to exactly how those activations from one layer bring about the activation in next layer it’s mean to be biological networks of neurons some groups of neurons firing cause certain other to fire. But the thing is that it is all done by the mathematical formula like Sigmoid function; it uses some functions to predict or recognize patterns and learn from this pattern.

Now take look at how Google uses the power of neural networks

The most visible developments in Google’s neural network research have been the DeepMind network, the “machine that dreams.”

What if computers could dream?

In fact, they can. Google’s groundbreaking DeepDream software is turning AI neural networks inside out to understand how computers think. The field of machine learning and artificial intelligence is growing by leaps and bounds. Already, it has enabled things that seemed like magic a few years ago Alexa can tell you jokes, AI chatbots can answer your questions on Facebook Messenger, and Google’s advanced neural-net machine learning algorithms can recognize pictures of virtually anything. It was while pondering this question that Google engineers had a crazy idea they decided to let the algorithm dream.

Now known as DeepDream, Google has taken their neural net image recognition software and programmed it to run backward. Artificial neural networks work by being trained on millions of examples and gradually adjusting network parameters until they result in the classification desired. For example, a net might be shown millions of pictures of dogs until it is able to recognize dogs with a high degree of accuracy. Due to the complexity of the learning process, however, what is actually happening inside the neural net is poorly understood. To understand this process better, Google engineers decided to turn the operation inside out.

Initially, engineers fed the neural net an image of a random scene and allowed it to look for an image to recognize. Once the program came back with a fit, they then instructed it to change the original image to fit the program’s classification algorithm better. Repeating this cycle many times resulted in psychedelic images “dreamed” by the algorithm essentially manifestations of what the algorithm understood the classification in question to be.

Taking it a step further, Google then tested their neural net on pure white noise. Given images with absolutely no patterns, this allowed the algorithm to dream much like a human does when looking at clouds. When we see a shape that is similar to something we recognize, we might say a cloud looks like a rabbit or a car. This abstract generalization is formed in the depths of our consciousness. By replicating this process with DeepDream, Google is gaining insight into how its algorithm thinks.

Images generated from random noise

In some cases, the results are surprising! According to a Google researcher, “The problem is that the knowledge gets baked into the network, rather than into us ”

Current products of google which use the power of Neural Networks

From smartphone assistants to image recognition and translation, a myriad of AI functionality hides within google apps that you daily use.

  • Google the search engine is powered by AI: According to Wired’s Cade Metz; Google’s search engine was always driven by algorithms that automatically generate a response to each query. But these algorithms amounted to a set of definite rules. Google engineers could readily change and refine these rules. And unlike neural nets, these algorithms didn’t learn on their own. But now, Google has incorporated deep learning into its search engine. And with its head of AI taking over the search, the company seems to believe this is the way forward.
  • Google Ads and Doubleclick both incorporate Smart Bidding which is a machine learning-powered automated bidding system.
  • Google Maps Driving Mode estimates where you are headed and helps you navigate without any commands.
  • Youtube Safe Content uses machine learning techniques to ensure that brands are not displayed next to offensive content.
  • Google Photos suggesting which photos you should share with friends.
  • Gmail Smart Reply suggesting replies that match your style and the email you received.
  • Google Drive Smart Scheduling suggests meeting schedules based on the user’s existing schedules and habits.
  • Google Calendar Quick Access feature predicts which files will be used improving performance and user experience.
  • Nest Cam Outdoor leveraged machine learning to set up an automated outdoor security camera as explained here.
  • Google Translate uses an artificial neural network called Google Neural Machine Translation (GNMT) to increase fluency and accuracy of translations.
  • Google News uses AI to understand the people, places, and things involved in a story as it evolves, organize them based on how they relate to one another as explained in Google Blog.
  • Google Assistant is a voice assistant for smartphones or wearables that can search online your flight status or the weather when you get there. Touch and hold the Home
  • Google News uses AI to understand the people, places, and things involved in a story as it evolves, organize them based on how they relate to one another as explained in Google Blog.
  • Google Assistant is a voice assistant for smartphones or wearables that can search online your flight status or the weather when you get there. Touch and hold the Home button and find your Google Photos, access your music playlists, and more. Both Siri and Google Assistant do a decent enough job of finding restaurants, bars, and another kind of businesses nearby, but Google’s app came out on top in our tests, not just on the places it returned, but on the interface: results are presented in a simple carousel and you can quickly jump to a Google Maps view. Also Google Assistant remembers what you’ve already said, speaks in foreign languages. Like Apple’s Siri, it is much more than an assistant, despite the name: it will read your poetry, tell you a joke, or play a game with you.
  • Google Home: You will be able to get hands-free help from your Assistant embedded in Google Home. Say “Ok Google” to get the morning news or manage your schedule.

Computer engineering student at silver oak university

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store