Deep Learning for Digital Devices – Davis Sawyer, Co-founder and Chief Product Officer, Deeplite – How Deep Learning and Artificial Intelligence Can Increase Efficiency and Productivity, and Bring Faster Power to the Palm of Our Hands

Davis Sawyer, co-founder and chief product officer at Deeplite (deeplite.ai), delivers a comprehensive and detailed analysis of the many ways in which deep learning is enabling AI to bring faster, more efficient technology to our low-power devices.

Sawyer's company, Deeplite, seeks to implement their patented technology to allow industries to maximize the productivity of deep learning systems and to enable AI for new opportunities on low-power devices. Deeplite's AI-driven optimizer accelerates deep neural network processing and makes it more energy-efficient. Their faster, leaner neural networks enable AI to be more accessible, easier to scale, and more economical as well.

From translating speech to text automatically, to powerful photo classification and recognition, AI has enhanced our daily lives in an increasingly data-driven world. However, as Sawyer describes, there are still many fundamental challenges that must be addressed to fully integrate AI into our lives for maximum productivity. One challenge Sawyer discusses is how to put sophisticated computation on low-end devices such as smartphones, to bring the tech to the edge of computing.

Deeplite's mission is to fully integrate the AI technology so smartphone users can get the most powerful tech and computational processing for whatever their business or industry endeavors may be. Sawyer states that you can make a world fit for AI or you can make AI fit into the world. And at Deeplite, Sawyer's team takes the approach that it is best to look under the hood so to speak, to look at the fundamental calculations to see if there can be improvements in order to do a better job of delivering solid results.

Deeplite is working to optimize deep learning models to assist companies that are using these models in their business structure. Through reinforcement learning, Sawyer's product essentially provides a means to compress models automatically for a more cost-effective, faster path to usable AI solutions for business. Sawyer discusses the many areas in which Deeplite's technology can increase productivity and efficiency.

Sawyer discusses robustness, privacy, and latency concerns in regard to advancing AI in business and daily life, and he details how autonomous vehicles are one industry area in which Deeplite is lending a hand. Deeplite's engine can produce optimized models that cut down AI computational time, which increases the efficiency significantly. With sometimes as many as six hundred million parameters and huge data sets, the problems, as Sawyer details, are sometimes scalability, and the need to create algorithms where understanding is not sacrificed, but is in fact enhanced.

Sawyer discusses informational resiliency in deep neural networks. He explains the process of stressing a network, forcing it to meet constraints, etc., putting environmental pressure on intelligent systems, and seeing how they successfully adapt. It is this kind of discovery that pushes Sawyer and his team at Deeplite to dig deeper into the methods of utilization for deep learning systems, to enable them to continue developing better tech products to advance the many ways we use and benefit from technology.