Asgent Announces Start of Sales for Next Generation Endpoint Solution, Deep InstinctTM,
Protecting Endpoints from Malware in Real-time Through Deep Learning
- Interoperability with existing antivirus and Microsoft Azure OMS management solution –
Harel-Hertz Investment House served as advisors to Deep Instinct in engineering Japan Entry
September 13, 2017 Asgent, Inc.(JASDAQ: 4288)
Asgent, Inc. (President & CEO: Takahiro Sugimoto, located in Chuo-ku, Tokyo), a pioneer in network security and operations management solutions, announces the signing of a distributor agreement with DEEP INSTINCT, Ltd (CEO: Guy Caspi, located in Israel), and the start of sales of the next generation endpoint anti-malware solution "Deep Instinct".
It is already a well-known fact that traditional signature based methods, based on discovered malware, are not sufficient to protect against the massive number of new types of malware that appear constantly today. Also, sandboxes alone are also reaching their limits, as advanced malware appears that knows when it is being run in a virtual environment by a sandbox to determine if it is a malware.
In this environment, detection and prevention systems utilizing AI (artificial intelligence) are gaining much attention as new technologies. Deep learning, in particular, is capable of automatically discovering characteristics based on big data, that humans would traditionally have to sample from data and provide to machine learning, and repeatedly learn from them. Therefore, the greater the data set, the higher the quality of the deep learning engine.
Deep Instinct is the world's first NGEPP (Next Generation EndPoint Protection) product in the security field to detect a threat and prevent it before any damage occurs, by deep learning used a neural network*1. Through repeated deep learning of millions of malware samples, it is possible to detect and block malware before it is executed on the endpoint, regardless of whether it is known or unknown malware. Working in collaboration with major antivirus products such as Windows Defender enables extremely high detection rates. Also, in addition to the Deep Instinct management module being used individually, through integration with Microsoft's cloud platform Microsoft Azure's management solution, Operations Management Suite (OMS), it is possible to collect Deep Instinct logs for each endpoint along with other Microsoft Azure security and operation logs and provide unified management.
*1 Neural Network: a mathematical model expressing the relationship between human brain neurons and their neural pathways. Through changing the bond strength of synapses by learning, it is possible to develop high problem solving abilities.
Ahead of the start of sales in Japan, Microsoft Japan, Inc. Executive Officer Cloud & Enterprise Business Department Head, Satoshi Asano, has the following comment.
"We deeply welcome the Deep Instinct product from Asgent, comprising of the D-Appliance etc. operating on Microsoft Azure that enables endpoint protection for customers, utilizing deep learning.
Through this the administrator can uniformly manage both network and endpoint security information, and use this for developing security measures. Also, as it will become possible for many organizations to have advanced malware detection using deep learning at a low cost, without the need for hardware assets, they will be able to gain even stronger security."
Deep Instinct is comprised of the D-Brain (research labs), D-Appliance (management module) and D-Client (client agent). Their roles and features are listed below.
D-Brain: DEEP INSTINCT's research laboratory.
- A high detection rate is enabled through repeated learning of millions of malware samples using deep learning.
- Learning results are reflected in the prediction module every three months.
D-Appliance: Management module. Distributes to endpoints and manages prediction module created by the D-Brain.
- Operates on both cloud platforms such as Microsoft Azure and as on premise solution. When using the cloud, can be implemented at low cost without the need for network configuration changes.
- Integrates with Microsoft Azure OMS to enable uniform management of D-Client logs.
D-Client: Client agent. Protects endpoint from zero-day and targeted attack threats etc.
- As opposed to general antivirus products that require frequent updates, the prediction module is distributed once every three months, so it can be used whether the endpoint is online or offline.
- Memory usage is extremely low, so user feels no drop in performance following installation.
- Interoperates with major antivirus products.
In pre-release performance testing performed at Keio University in Japan, the detection rate for known malware samples was more than average 98.7%, and ransomware had a detection rate of 99.2%. It also achieved an average 80% detection rate on malware samples created by the university that were equivalent to unknown malware. Professor Osamu Nakamura and Appointed Professor Kazuma Kobayashi of the Faculty of Environment and Information Studies, Policy and Media Research Graduate Course at Keio University, who carried out the performance tests, have the following comment.
"As part of testing the detection rate of Deep Instinct, we use about 12,000 known malware samples, and malware samples modified from the 12,000 known samples (essentially unknown malware). As an initial result for an endpoint solution with an AI engine, the detection rate is satisfyingly high. We can expect even higher detection rates through future deep learning."
Moreover, Asgent plans on releasing test results from the official Japan release version in November, with the continued cooperation of Keio University.
Asgent will start sales of Deep Instinct (including Microsoft Azure support) with the Japanese localized D-Client from November 2017, and aims for 700 Million Yen in the initial year of sales.