Deep learning (DL), a transformative branch of machine learning and more broadly artificial intelligence (AI), is poised to transform every business segment and industry. AI Innovations are unfolding at an unprecedented pace, with new capabilities for language processing, image recognition, recommendation systems and many more rapidly evolving. Breakthroughs in DL hardware and software, as well as a massive expansion in DL-based capabilities and solutions from the edge to the cloud will continue over the next several years. In parallel, AI needs to evolve and expand from current DL approaches, adding new levels of cognitive capabilities to bring the technology that much closer to human intelligence in defined areas.
I have been part of the journey to increase machine intelligence for almost a decade now. After previous decades focused on several leading compute architectures and products at Intel, it became clear around 2010 that emerging AI capabilities would shape not only the future of compute, but also the broader concept of interaction between humans and machines. In 2012, we founded a Convolutional Neural Networks (CNNs) Lab to help evolve the technology roadmap for DL inference acceleration, natural language processing (NLP) and visual recognition. Over the last several years, Intel invested in integrating these DL capabilities into the CPU and other products, creating world-class technologies for dedicated DL acceleration, and restructuring the software stacks to deliver peak performance and programmability to fuel adoption and further innovation in this field.
While the majority of efforts in the industry focus on improving the efficiency and reach of DL technologies and products entering the market today, it is becoming clear that DL approaches have their limits and AI needs new innovations to acquire more human-like cognitive competencies. New technologies are needed to address real-world challenges like explainability, reasoning, out-of-distribution (OOD), extensibility to new domains with minimal additional data, system adaptation through continuous learning, and other limitations of current DL approaches.
To address these challenges, we have established Cognitive Computing Research at Intel Labs to drive Intel’s innovation at the intersection of machine intelligence and cognition. Our efforts combine the latest in deep learning, with the integration of knowledge structures as well as logic-based reasoning and neuro-symbolic AI, in order to build self-learning AI that can make informed decisions in complex context-rich situations. We will draw on the cutting-edge research from related Intel Labs teams including brain-inspired computing, visual understanding, neuromorphic computing, probabilistic, NLP and more, working together to drive the industry forward.
Deep learning makes AI systems incredibly effective in recognition, perception, translation and recommendation system tasks. The nascent technologies for the next wave of machine learning and AI will create an entirely new class of AI solutions with higher understanding and cognition. We look forward to building next-generation AI systems that will one day understand this blog post and other informative content – and deliver even greater benefits to our lives.
Gadi Singer is Vice President of the Intel Labs, Director of Cognitive Computing Research.