Make Smarter, Faster Decisions from Ever-Growing Sets of Data
Many business problems can be maddeningly complex, involving the analysis of extremely large and diverse data sets. Organizations often face questions like "Can IT build a spam-hunting application to stop junk postings from choking our company’s community web forums?"
Answering and acting on such complicated, data-intensive questions might exceed the ability of traditional business intelligence (BI) and rule-based analytics systems. These approaches may not be forward-looking or flexible enough in dynamic business environments deluged by Big Data from the cloud, social media, smart mobile devices and the Internet of Things (IoT). To better understand and solve fast-changing challenges involving enormous data troves, companies in every industry – from healthcare to banking, transportation to manufacturing, education to retail and more – are upgrading their analytics capabilities with machine learning.
A subset of artificial intelligence (AI), machine learning uses specialized software algorithms that iteratively “learn” and adapt as programs sift through massive data sets. These examples allow the organization to discover and act on patterns, insights and trends. And that produces better results over time without human intervention. These benefits are making machine learning more mainstream every day. Computers that learn drive a wide array of real-world applications: IoT data analysis, computer server monitoring, targeted advertising, image recognition, route scheduling, genetic sequencing, gaming, autonomous vehicles, energy exploration, facial recognition and many more.
Machine learning produces data-driven insights and complex, actionable decisions from extremely large data sets much more quickly and reliably than human analysis, traditional BI or other AI approaches. Machine learning drives greater efficiencies in business operations, improves security and sparks data-powered innovation with new products and services tailored to customer behaviors.
Create Smarter Internal Business Processes
According to a recent study conducted by Bain & Company, companies that use machine learning and analytics are twice as likely to make data-driven decisions, five times as likely to make decisions faster than their competitors, three times as likely to execute more quickly on those decisions and are twice as likely to have top-quartile financial results. For many organizations, ascending the business intelligence maturity curve with machine learning begins with using machine learning to improve key internal business processes.
Some high-level examples include:
Improved hiring and worker performance : A global fast food company uses machine learning to gain insight on talent acquisition, retention and employee performance. Such “people analytics” provides deep insights about HR data using predictive modeling techniques on multiple, integrated data sources.
Customized Marketing: A major Italian bank created a cognitive analysis system to analyze customer data and find hidden hotspots of opportunity. The approach produced a targeted outbound marketing program that significantly improved conversions.
Customized Price Quotes: A leading global software company uses a computerized price quoting driven by machine learning to tailor precise, targeted options for every customer and prospect. The company has more accurate, on-target predictions because machine learning integrates with customer relationship management (CRM) and enterprise resource planning (ERP) systems.
Personalized Medicine: Growing numbers of healthcare providers use machine learning to power a data-driven, precision-medicine approach that identifies the most cost-effective, personalized treatment options.
For many early adopters in healthcare and elsewhere, machine learning is reshaping their businesses through improved efficiencies, new discoveries, improved products and services or better customer experiences.
1. Can IT build a spam-hunting application to stop junk postings?
One of machine learning’s core strengths – the ability to discern unusual patterns within vast data pools under rapidly changing conditions – makes the technology well-suited for faster detection and mitigation in the realm of security. For example, machine-learning algorithms look for patterns in how cloud data is accessed and report anomalies that can predict security breaches. Payment processors use learning algorithms to track credit and debit card users purchasing patterns, flagging anomalies such as unusual purchase amounts, or interactions with merchants or in geographic locations that point to possible fraud.
Intel Corporation, for instance, uses machine learning techniques to study and stop junk messages on website community forums that serve customers, partners and workers. One of Intel’s larger forums was inundated by as many as 10,000 spam postings per day.
The usual remedy, enlisting volunteer moderators to delete junk posts, didn’t scale and consumed too much staff time. It also wasn’t possible to use filters stop the problem. And the ubiquitous growth of new spam-bots made it ever harder to define a universal rule blocking a particular word or phrase. Was the message promoting offshore casino gambling or was it from one of Intel’s entertainment industry customers?
Frustrated, Intel’s IT group came to another solution. The company already was putting automation to work in many areas, such as PC health monitoring and factory processes. Since automation had increased efficiency and effectiveness in those areas, why not use machine learning for autonomic spam control? Using sophisticated machine-learning techniques, Intel engineers built a spam-filtering service that automatically blocks unwanted and malicious messages. Text analytics let the system detect profanity and objectionable content in 75 languages. And reputation-engine monitors user profiles to discern the likelihood of a given source submitting spam.
Attacks dropped off immediately after Intel implemented the program, and spam levels have remained manageable ever since. Spikes in junk posts have all but disappeared, thanks to the service’s ability to learn dynamically and block unwanted messages.
2. How can we make smarter use of sensors?
For many businesses, the most valuable use of machine learning is making sense of and exploiting the torrents of data from trillions of sensors and other devices connected to the Internet of Things (IoT) and Industrial Internet of Things (IIoT). In the past year, auto and tool makers, pharmaceutical firms, fleet operators and companies in other industries have begun or expanded the use of machine learning and analytics in IoT as a foundation of autonomous manufacturing.
For instance, Siemens AG, as a first step in building an autonomous manufacturing plant, created a cloud-based, open IoT ecosystem called MindSphere*. The robust digital platform captures, stores and analyzes data generated by manufacturing control systems and sensors on equipment connected via the IoT1. Siemens uses machine learning to study this data and to analyze the entire supply chain. In this way, the international industrial giant determines where to make improvements on the manufacturing line that result in the biggest gains for the company. This “smart data” gives Siemens managers actionable insights that improve equipment uptime and increase the efficiency of production operations.
Machine learning and analytics is the foundation of autonomous manufacturing where virtually all processes eventually will be digitally based and highly automated. The rapid maturation of learning algorithms has given manufacturers the ability to collect, store and analyze huge amounts of data in real time and to turn that data into actionable sets of information. More importantly, machine learning helps companies get smarter by adding proactive devices with dynamic machine learning of their environment, users and history to assist in analysts’ operating decisions.
3. Can we aggregate the financial histories of many people to help detect fraud?
Improving current business processes is only the first step in leveraging the power of machine learning. The never-before-possible insights that the approach produces can inspire new products, services and new ways of doing business. It can transform entire industries.
Consider the retail sector. Brick and mortar stores are constantly working to re-invent themselves as they compete against each other and against online rivals. Experts say success – and survival – in retail depends on creating a highly personalized mix of online and in-store selling. One retail company recognized that knowing what customers are interested in — and getting those products in front of those customers at the right time — removes a lot of friction from the sales process. That requires a mix of connected devices and analytics software.2
The company created an application, which uses in-store sensors and machine learning to capture and analyze style preferences and buying trends to guide clothing designers.
Another of the retailer’s applications helps add options to the outfits customers already bought by suggesting other items based on the initial purpose. If a consumer buys a shirt from the brand’s e-commerce site, the AI-based platform presents a menu of additional garments and accessories to “complete the look.” This approach drastically improves revenue at the point of sale. Indeed, the retail company currently delivers 4.5 million recommendations to customers every day. Both the sensor and option-fitting applications give the retailer a wealth of valuable information about customer preferences and behaviors that are used to personalize and improve sales and service.
With leading retailers shifting their focus to machine learning and analytics, it's clear that data is becoming a driving force as the entire industry strives to transform itself. The trend likely will grow, as retailers such as Lowe’s* Home Improvement and Amazon* Go grocery stores deploy sensor-laden robots that conduct real-time shelf audits to optimize inventory. The retail example demonstrates the opportunities presented by machine learning. The difference between the companies that embrace the technology and those that do not will determine which become industry disruptors and which fail and become the disrupted.
How Intel Helps Realize Machine-Learning Benefits
Machine learning and data analytics help make organizations smarter, faster, more efficient and more innovative. Better, faster, real-time decisions drive operational improvement and new products and business models that provide a genuine competitive edge. Whether a business is just getting started with a machine learning pilot or has taken on more ambitious and advanced analytics projects, Intel provides a variety of resources and technologies that help create the robust, end-to-end architecture that machine-learning technologies require.
Intel’s performance-optimized portfolio and rich solution ecosystem supports the progression to advanced machine learning analytics. Intel does this by partnering with top system integrators and technology vendors that provide the framework for the distributed storage and processing of big data.
The company’s experts and experienced partners help organizations make the best decisions on a wide range of machine-learning technology and implementation needs.
The bottom line is that machine learning in analytics has arrived. Intel has proven the value of machine learning for both its own business and for its customers and partners.