Businesses are increasingly seeing the need for simpler, faster ways to harness large data sets and extract useful insights at scale. Thanks to advancements in artificial intelligence (AI), we are moving from experimentation to production. Companies from General Electric to Walmart are using AI and analytics to recognize the power of the data they and their customers generate. The business benefits can be massive, but implementation is a huge and complex undertaking that requires an end-to-end data analytics pipeline.
At the O’Reilly AI Conference in Beijing, Intel showed its continued commitment to solving real business problems across multiple market segments with solutions built on modern, scalable hardware and software architectures. A unified, Intel-based architecture also provides IT staff with the ability to harness familiar software and advanced toolkits that support business growth without requiring excessive investments in single-purpose products. And since there’s no need to move data across different environments, companies can help ensure greater data privacy.
Intel Continues Its Support for Open Source in AI and Analytics
Analytics and AI architecture built with popular open source software has the advantages of greater transparency and the ability to harness the innovation of the larger open source community. Open source technologies also free IT staff to create optimized, end-to-end innovative solutions that can quickly and cost-effectively scale because developers can see the source code to create customizations ideally suited to their situation.
Intel’s history of contributing to open source projects extends back decades, and we’re taking that same view with our work in data analytics and AI today. Our goal is to enable a robust and vibrant ecosystem by optimizing the most popular frameworks and abstractions that make the best use of Intel hardware across training and inference. Learn more about Intel’s exciting new efforts around the open source Analytics Zoo toolkit for TensorFlow, Keras, and Apache Spark in my colleague Jason Dai’s blog.
To deploy AI and analytics in a growing organization while maintaining effective TCO, data scientists and developers will need to run deep learning training across many servers (or nodes), a strategy known as distributed training. The open source software toolkit Analytics Zoo is designed to scale in a near-linear fashion across cores and nodes to dramatically reduce the time to train models on Intel’s latest platforms featuring 2nd generation Intel® Xeon® Scalable processors and Intel® Optane™ DC persistent memory.
Scaling in China and Across the World
To further help meet the demands of today’s data-rich markets, I am happy to support the launch of the Intel Virtual Innovation Center for Data Analytics and AI. Based in China, the virtual center connects Intel experts with technology partners and customers to develop end-to-end, integrated solutions at scale. The center will provide access to the latest Intel data-centric advancements in hardware platform technologies, as well as related optimized libraries, software, and tools. In addition to enabling customer use cases, Intel will provide training resources to developers and drive a new research pipeline with academics to accelerate technical innovation.
The business benefits of AI and analytics are already well-documented across several industries. As new technologies are developed and current technologies are honed, companies that have a clear path to take advantage of the data they and their customers generate will have a dramatic edge up on competitors. Having a unified platform optimized for the workloads IT professionals know and trust is the best way to keep TCO low and be prepared for changes the future will inevitably bring, no matter where your company operates across the globe.
For more information on the data analytics and AI technologies that support the Virtual Innovation Center, visit http://software.intel.com/AIonBigData. Learn more about how advanced analytics is helping reshape global business at intel.com/analytics.