accelerate-intelligent-solutions-with-machine-learning-platform-brief
- 1. Authors Nidhi Chappell Director, Machine Learning Datacenter Group Herbert Cornelius Principal HPC Solutions Architect, Influencer Sales Group Machine learning platforms powered by Intel® technology help transform data into actionable business intelligence through accelerated model training, fast scoring, and highly scalable infrastructure Accelerate Intelligent Solutions with a Machine Learning Platform Executive Summary Machine learning enables businesses and organizations to discover insights previously hidden within their data. Whether exploring oil reserves, improving the safety of automobiles, or mapping genomes, machine learning algorithms are at the heart of innovation and business intelligence. Unleashing the power of machine learning, however, requires certain ingredients: access to large amounts of diverse data and the right skill sets, optimized data platforms, powerful data analysis tools, and a highly scalable and flexible compute and storage infrastructure. Intel’s high-performance computing (HPC) reference architectures are optimized for machine learning. Built on a hardware foundation that incudes compute, memory, storage, and network, these platforms include an optimized, scalable software stack for predictive analytics. By using a machine learning platform based on Intel® architecture, businesses can gain scalability, effectiveness, efficiency, and lower total cost of ownership (TCO) while reducing time to market for intelligent solutions that can give them a competitive market edge. Solution Brief Data Center High-Performance Computing Figure 1. Insights are there, but they lie buried in huge volumes of data. Machine learning can help companies uncover those insights, which they can use to develop innovative, intelligent solutions. Deep Insights Provide Competitive Edge Machine Learning with Smarter Algorithms Processes overwhelming volumes of data 10011111010001011010101100111110100010110101011001111101000101101010110011111010001110100010110101011001111101000101101 100111110100010110101011001111101000101101010110011111010001011010101100111110100010110101011001111101101 10011111010001011010101100111110100010110101011001111101000101101010110011111010001011010101100111110100010 1001111101000101101010110011111010001011010101100111110100010110101011001111101000101101010 1001111101000101101010110011111010001011010101100111110100010110101011001111101000101101010110011111010001011 10011111010001011010101100111110100010110101011001111101000101101010110011111010001011010 100111110100010110101011001111101000101101010110011111010001011010101100111110100010110101011001111101000101101010110011111010001011010101 1001111101000101101010110011111010001011010101100111110100010110101011001111101010001011010101 10011111010001011010101100111110100010110101011001111101000101101010110011111010001011010101100 100111110100010110101011001111101000101101010110011111010001011010101100111110100010110101011001111101000101101010110011111010001011010101 10011111010001011010101100111110100010110101011001111101000101101010110011111010001011010101100110001011010101 100111110100010110101011001111101000101101010110011111010001011010101100111110100010110101011001111101000101101010111
- 2. Solution Brief | Accelerate Intelligent Solutions with a Machine Learning Platform 2 Business Challenge: Using Machine Learning Effectively Businesses in every industry can gain a competitive advantage and generate new revenue by delivering intelligent products and services that are more personalized, efficient, and adaptive. But CIOs are buried in data—the challenge lies in effectively using machine learning techniques (Figure 2) such as deep learning, computational statistics, mathematical optimization, and artificial neural networks to build intelligence into solutions. Machine learning – an outgrowth of artificial intelligence – enables researchers, data scientists, engineers, and analysts to automate analytical model building by constructing algorithms that can learn from and make predictions based on data. The explosion of big data has made machine learning an important differentiating factor across many industries. For example, bioinformatics’ high-throughput techniques can rapidly produce terabytes of data that overwhelm conventional biological analysis. Ultra-scalable, high-performance machine learning platforms, however, can quickly process vast amounts of data. Machine learning also has applications in the areas of modeling web browsing behavior, spam filtering, optical character recognition, and fraud detection, just to name a few. However, the powerful potential of machine learning seems out of reach for many organizations. Using machine learning technologies effectively can be challenging. To be successful, the following elements are necessary: • Access to large amounts of diverse data • Optimized data and compute platforms to manage and process data • Powerful data analysis software to build sophisticated predictive models • A highly scalable, flexible infrastructure (compute, memory and storage, and network) on which to develop, train, and deploy models based on machine learning • A pool of appropriately skilled talent, such as data scientists and solution developers, that can efficiently manage insights from data Machine Learning Use Cases Span Multiple Industries Whether an organization is developing models for disease prevention or storm prediction, machine learning can speed results while delivering a higher degree of accuracy. Retailers can better predict customer purchases and reduce customer churn by delivering targeted offers; utilities can more accurately forecast and prevent potential outages; and companies can better automate help desk services and improve customer service. For example, in Europe, more than a dozen banks have replaced older statistical-modeling approaches with machine learning techniques and, in some cases, experienced 10-percent increases in sales of new products, 20-percent savings in capital expenditures, 20-percent increases in cash collections, and 20-percent declines in churn.1 Figure 2. Various machine learning techniques pose unique challenges, but they require common elements of data, skill sets, and the right platform components. Machine Learning Data Processing Curation Identify sources and understand relationships Variety is massive, continuously new sources Training Train an algorithm to build a model Model build time is critical Scoring Deploy models for classification, prediction, and recognition of new data Requires easy distribution, sensitive throughput, and TCO at scale Deep Learning Technique • Many hidden layers • Features are learned • Complex data Other Techniques Techniques include: computational statistics, mathematical optimization, and artificial neural networks • Clustering, regression, and classification using one or two hidden layers • Features are engineered Using Machine Learning Techniques Effectively “Humans can typically create one or two good models a week; machine learning can create thousands of models a week.”2 —Thomas H. Davenport Analytics Thought Leader (excerpt from The Wall Street Journal)
- 3. Solution Brief | Accelerate Intelligent Solutions with a Machine Learning Platform 3 Solution Value: New Insights Enable Better Business Decisions Intel’s high-performance computing (HPC) machine learning reference architectures offer the following enterprise benefits: • Shorter time to train models, with scalable multi-node configuration for complex neural networks • High throughput scoring on standard, energy-efficient server-class infrastructure • A single architecture for multiple advanced analytics requirements Organizations can build accurate models faster and deploy intelligent solutions quickly, while decreasing total cost of ownership (TCO). Intel’s portfolio of innovative technologies – both hardware and software – provide balance, portability, and high performance through tight, system-level integration and modernized code. Intel’s optimized HPC technologies lay the foundation for a holistic machine learning platform. This platform is built on industry-standard hardware and can be deployed on‑premises or in public or hybrid cloud environments. The platform can scale from small clusters to supercomputers. Intel’s ongoing investment in HPC technologies opens the path for new parallel, neural, and quantum computing options. With attractive performance and TCO provided by platforms based on Intel’s machine learning reference architectures, companies can optimize value from their data through advanced analytics. Solution Architecture: Fully Optimized Machine Learning Environment Intel’s machine learning reference architectures help companies build platforms that can tap into the power of machine learning. An Intel® architecture-optimized infrastructure serves as the foundation for these platforms, which are ideally suited for a broad range of machine learning workloads. • Compute. Servers equipped with Intel® Xeon® processors help keep costs affordable while delivering exceptional performance, agility, reliability, and security. Intel® Xeon Phi™ processors and coprocessors offer highly parallel performance and can scale to over 100 software threads, make extensive use of vectors, and efficiently use local memory bandwidth—a benefit for the iterative nature of machine learning workloads. Other technologies include built-in field-programmable gate array (FPGA) modules for augmented specific acceleration. • Memory and storage. As model sizes increase, it is important to keep data close to memory to reduce latency while processing large data sets. Example technologies include 3D XPoint™ technology, Intel® Optane™ technology, and rugged, high-performance PCI Express*- and non- volatile memory Express-based solid-state drives (SSDs). • Network. Effective machine learning platforms require a high-performance, low-latency fabric like Intel Omni-Path Architecture to maximize memory capacity and floating-point performance and accelerate results. Example technologies include highly scalable Intel® Omni-Path architecture and Intel® Ethernet Server Adapters (10 GbE and 40 GbE). As shown in Figure 3, scalable data and analytics platforms are layered on this infrastructure, which can then efficiently run individual analytics applications. fASterDeCiSionS3 Companies that use analytics are: • 5x more likely to make “much faster” decisions than competition • 2x more likely to have top-quartile financial performance • 3x more likely to execute decisions as intended • 2x more likely to frequently use data when making decisions Figure 3. A fully optimized machine learning environment is built on tightly integrated Intel® technologies for accelerated insight discovery at a lower cost of ownership. Applications Analytics-powered vertical and horizontal solutions Machine Learning Frameworks and Algorithms Multi-layered, fully optimized algorithms • Intel® Math Kernel Library • Intel® Data Analytics Acceleration Library Performance and Security Silicon and software enhancements to protect and accelerate data and analytics Trusted Analytics Platform Open-source platform for collaborative data science and analytics app development Data Open-source, Hadoop-centric platform for distributed and scalable storage and processing Infrastructure Optimized for Intel® Architecture Software-defined storage, virtualized compute, networking, and cloud Scalable Data and Analytics Platforms
- 4. Solution Brief | Accelerate Intelligent Solutions with a Machine Learning Platform 4 1 McKinsey Quarterly, June 2015, “An executive’s guide to machine learning,” http://www.mckinsey.com/industries/high-tech/our-insights/an-executives-guide-to-machine-learning 2 Source: http://www.sas.com/en_us/insights/analytics/machine-learning.html 3 Bain and Company, September 11, 2013, “Big Data: The organizational challenge,” http://www.bain.com/publications/articles/big_data_the_organizational_challenge.aspx Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more complete information visit intel.com/performance. Cost reduction scenarios described are intended as examples of how a given Intel-based product, in the specified circumstances and configurations, may affect future costs and provide cost savings. Circumstances will vary. Intel does not guarantee any costs or cost reduction. Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. No computer system can be absolutely secure. Check with your system manufacturer or retailer or learn more at intel.com. No license (express or implied, by estoppel or otherwise) to any intellectual property rights is granted by this document . Copyright © 2016 Intel Corporation. All rights reserved. Intel, the Intel logo, Xeon, Xeon Phi, 3D XPoint, and Optane are trademarks of Intel Corporation in the U.S. and/or other countries. * Other names and brands may be claimed as the property of others. 0616/RMCN/KC/PDF Please Recycle 334076-001US Each layer is optimized to provide extremely high performance: • Machine learning frameworks like Spark MLlib*, Caffe*, Theano*, Torch*, and CNTK* can scale across distributed systems based on Apache Spark* and MPI (Message-Passing Interface). Through code modernization, these frameworks are optimized to take advantage of parallelization at the thread, data, and vector levels, as well as the innovative memory/storage hierarchy. • Machine learning algorithms can extract exceptional performance from the underlying hardware platform using mathematical building blocks from the Intel® Math Kernel Library and the Intel® Data Analytics Acceleration Library, which are tuned for high performance on Intel® processors. • The performance and security of data acquisition, data management, and data processing engines are enhanced by open source products such as Apache Spark, Apache Hadoop*, and the Lustre* file system as well as optimized proprietary software products. Intel invests in building the broadest ecosystem possible and works closely with fellow travelers including academic researchers, OEMs, communications service providers (CSPs), ISVs, and system integrators. The result is machine learning technologies customized for your solution. Conclusion Machine learning is becoming a business necessity; it can help organizations quickly build models that enable development of intelligent solutions—which in turn create new revenue streams and differentiate those organizations from their competitors. But whether exploring underground oil reserves or improving the safety of automobiles, organizations need a highly scalable, balanced and robust infrastructure for HPC that speeds discovery and innovation and decreases time to market. Platforms based on Intel’s machine learning reference architectures and HPC solutions help enable organizations to fully realize the benefits of machine learning, while increasing performance and decreasing TCO. Intel is continuing to invest in HPC technology to help organizations meet the machine learning needs of tomorrow, driven by the explosion of data and the Internet of Things. Find the solution that’s right for your organization. Contact your Intel representative, register at Intel IT Center or visit intel.com/machinelearning. Solutions Proven By Your Peers Intel Solution Architects are technology experts who work with the world’s largest and most successful companies to design business solutions that solve pressing business challenges. These solutions are based on real-world experience gathered from customers who have successfully tested, piloted, and/or deployed these solutions in specific business use cases. Solution architects and technology experts for this solution brief are listed on the front cover.