- Deep Instinct is applying end-to-end deep learning to cybersecurity.
- It is an approach that allows it to predict and prevent cyberattacks across a company’s network.
- Deep Instinct wants to turn that same data into an enterprise’s greatest defensive asset.
- Deep Instinct creates a standalone neural network that the organizations can deploy.
The increasingly rich data companies are collecting makes them a more tantalizing target for attacks. But Deep Instinct wants to turn that same data into an enterprise’s greatest defensive asset.
Deep Instinct is applying end-to-end deep learning to cybersecurity, an approach that allows it to predict and prevent cyberattacks across a company’s network, according to CEO, Guy Casp
Today, Deep Instinct announced it has raised $100 million in a round led by BlackRock. Other investors include Untitled Investments, The Tudor Group, Anne Wojcicki, Millennium, Unbound, and Coatue Management. The company has now raised a total of $200 million.
Deep Instinct’s for security
The New York-based company is part of a growing wave of startups turning to machine learning and artificial intelligence to combat the rising number of cyberattacks. The industry is optimistic that this ability to automate defenses will help companies gain an edge against increasing sophistication and well-funded hackers.
But Deep Instinct is trying to go a step beyond the way others are using AI and machine learning for security. The company creates deep neural networks. These networks allow it to avoid using feature processing. This adds an additional step and slow reaction time.
With traditional machine learning, Caspi explained, executable files cannot undergo processing directly. Instead, they must undergo conversion into a list of features. These features are then fed into a machine learning model.
How Deep instinct’s works
Deep Instinct’s end-to-end deep learning system uses the raw data as input without needing to convert it. The company trains its model in its own labs, rather than on the customer’s premises, by feeding it hundreds of millions of malicious and legitimate files. This huge-scale training workload relies on Nvidia GPUs.
Once the training is finished, Deep Instinct creates a standalone neural network that can be deployed to an organization, where it starts protecting every device connected to the network. Because the system doesn’t require agents, it can be rapidly installed, including covering applications currently running. And it can recognize previously unknown types of attacks without needing to be constantly updated.
As a result, Deep Instinct claims it can identify and stop attacks within 20 milliseconds while reducing false positives by 99%.
Caspi said he wants to use the latest funding to accelerate growth with an eye toward an IPO in the next couple of years. For now, that means ramping up sales and marketing. The company plans to reserve about 30% of the money for product development.