DeepCube Launches Deep Learning Software Accelerator to Enable Real-World AI Deployments
TEL AVIV, Israel, May 12, 2020 -- DeepCube announced the launch of the only software-based inference accelerator that drastically improves deep learning performance on any existing hardware.
Today, deep learning deployments are very limited and are primarily optimized for the cloud; and, even in these cases, they incur extensive processing costs, significant memory requirements, and expensive power costs, due to intensive computing demands. These challenges also plague deep learning deployments on edge devices, including drones, mobile devices, security cameras, agricultural robots, medical diagnostic tools and more, where the current size and speed of deep neural networks has limited their potential.
DeepCube focuses on research and development of deep learning technologies that improve the real-world deployment of AI systems. The company’s numerous patented innovations include methods for faster and more accurate training of deep learning models and drastically improved inference performance on intelligent edge devices. DeepCube’s proprietary framework can be deployed on top of any existing hardware (CPU, GPU, ASIC) in both datacenters and edge devices, enabling over 10x speed improvement and memory reduction.
“Many deep learning frameworks were developed by researchers, for researchers, and are not applicable to commercial deployment, as they are hindered by technological limitations and high cost requirements for real-world applications,” said Dr. Eli David, Co-Founder, DeepCube. “DeepCube’s technology can enable true deep learning capabilities within autonomous cars, agricultural machines, drones, and could even help potentially monitor for and prevent future global health crises, much like the one we are facing now in 2020.”
Inspired by the way the human brain develops during early childhood, DeepCube’s patented technology continuously restructures and sparsifies deep learning models during the training phase to maintain high accuracy and greatly reduce the size of the AI model. Several of the largest semiconductor companies are already using DeepCube’s technology to drive significant increases in speed and memory reductions for deep learning inference with minimal drops in accuracy.
DeepCube is co-founded by Dr. Eli David and Yaron Eitan, who bring decades of experience not only within deep learning research and real-world execution, but also in technology entrepreneurship. Dr. Eli David is a leading deep learning expert and has published over fifty papers in leading artificial intelligence journals and conferences, focusing primarily on applications of deep learning and genetic algorithms in various real-world domains. Prior to DeepCube, Dr. David co-founded Deep Instinct, a World Economic Forum Technology Pioneer, and the first company to apply deep learning to cybersecurity. Yaron Eitan is a serial tech entrepreneur and investor with over thirty years of experience that he will apply to guide DeepCube’s business execution.
“The initial improvements we’ve generated through our POCs demonstrate that this technology can enable true deep learning capabilities across the entire AI deployment market, in any sector or industry, which will be critical as AI continues to penetrate new markets and vastly improve processes in industries like the medical field that need it most,” added Yaron Eitan, Co-Founder, DeepCube.
DeepCube has already been granted four patents in addition to filing for four others thus far.
To learn more about DeepCube’s industry-first technology, visit www.deepcube.com.
DeepCube is an award-winning deep learning pioneer that provides the industry’s first software-based inference accelerator to drastically improve deep learning performance on existing hardware. Modeled after the way the human brain develops during childhood, DeepCube’s patented technology is the first to be purpose-built for deployment of deep learning models on data centers and intelligent edge devices. Its proprietary framework can be deployed on top of any existing hardware, resulting in drastic speed improvement and memory reduction.