Google’s DeepMind artificial intelligence startup has a new tool for analyzing the behavior of its neural networks, and it has developed an artificial neural network that it hopes will help it predict the behavior and predict future events of humans.
“Our current method of learning is extremely inefficient and limited, with limited data to draw from, and the results are often ambiguous,” said DeepMind’s chief scientific officer, Dr. Natacha Pishevar.
“The DeepMind neural network we’re developing is an algorithm that can learn by analyzing data, and we are excited to see what it can do.”
It was recently named as one of the top 10 most promising AI startups by venture capital firm Andreessen Horowitz, and its founders and CEO, Sebastian Thrun, are currently working on a second generation of its AI.
The algorithm can identify patterns in data, predict when certain behaviors will occur and predict what people might do if given certain scenarios.
For example, it can predict what a person might do in a situation where they are likely to be in the company of other humans, or in a scenario where a person is likely to take the company’s products to the next level.
The neural network can also learn from context and context clues in an environment, as well as use the data to build its own model of how humans will react to a situation, said Pishever.
“In a scenario that the human is in, the algorithm can learn how to interpret the data, based on what the context and the context clues are,” said Pishel.
“We can build our own machine learning model that learns and adapts to new data.”
The AI system was developed by DeepMind in partnership with the University of California, Berkeley, and other academic institutions.
The software can learn from previous data and learn from new data, which allows it to create a model that can predict the actions of humans that could be the subject of future research.
Pishev also told The Huffington Googles AI research team will be working on further improvements in the next year or two.
“It will be a good year for us to go forward with this program, because we’ve had a lot of successes and we’ll be able to build on that success,” Pisheve said.
The program, called DeepMind Vision, is the first of its kind to use neural networks to predict what will happen if certain events take place in a given scenario.
It can also help with other applications, such as the predictive modeling of the behavior that is used in the health and wellness industries, and artificial intelligence in other fields.
Pishel said the team has developed the algorithm so that it can also predict the responses of people who are in a position to be part of a specific team.
“For example, the model that is trained to predict whether someone will be part the company or not, is used to create the business relationship,” Pishel explained.
“If the AI system learns from this data, it could predict which of these relationships the AI will create based on how it will behave in the future.”
Pisheval is the co-founder of the Artificial Intelligence Research Lab at the University Of Chicago.