Machine Learning: where did it start, and where is it going? The idea of algorithms has existed for a long time, but the modern concept of Machine Learning has its roots in 1950s computer science. With the strides made in data science and computer manufacturing—not to mention storage and data management—Machine Learning is becoming a great enabler of IT and business operations worldwide.
The History of Machine Learning
Is Machine Learning just another algorithm? The answer is a definitive “yes and no.” While the concept of algorithms has been around for thousands of years (thanks to Persian mathematician Muhammad ibn Mūsā al- Khwārizmī ), modern computational algorithms and the advent of Machine Learning (ML) has taken decades to evolve. Arthur Samuel is credited for coining the term “Machine Learning” while working at IBM, using the game of checkers in his research. Famed checkers master, Robert Nealey, lost a checkers game against an IBM 7094 computer in 1962. That famous match is still considered a major milestone in computer science.
Since then, the technological developments in storage and processing power have paved the way for modern products that we’re all familiar with, such as Netflix’s recommendation engine or personal assistants like Siri and Alexa. ML is typically considered to be a subset of artificial intelligence (AI) that enables computers to “learn” without being explicitly programmed with predefined rules. It focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy.
ML is an advanced algorithm (or model) that learns patterns in data and then calculates similar patterns in new data. So instead of setting up precise rules for an item you want to classify, for example, you can use hundreds of rules of an item. The computer finds the patterns in these descriptions and uses that pattern to identify and categorize the items.
ML concentrates on developing computer programs that can grow and change when exposed to new data. This reactive capacity, combined with our current ability to process massive amounts of data, means that ML can handle complex actions with efficiency and accuracy.
Machine Learning Gathers Steam with Big Data
The opportunities to use ML in myriad situations have grown exponentially in concert with computational power, device design, and data access and storage. From playing a simple game of checkers on a mainframe computer as large as your living room to shopping, banking, and video conferencing from a device that fits in the palm of your hand, ML has impacted business, education, and home life in equal measures. Today, ITOps organizations can use ML in various ways to support the business goals, particularly for companies that deal with volumes of data, media, documents, search needs, forecasting inventory and resources, and fraud detection.
For all these applications for Big Data, ITOps teams continue to be responsible for:
- Data backups and restorations, server management and configuration, optimizing the performance and resource allocation for effective delivery
- Management of overall tech stack, including IT infrastructure encompassing computation, networks, and hardware as well as the software applications plus configuration of any components
- Mitigation of disaster, including security implementation and management, along with disaster planning and management.
Large cloud computing companies—including Microsoft, AWS, and Google—as well as specialized tech firms like Databricks, MathWorks, and InData Labs, offer ML solutions to help optimize business operations, improve customer experience, and accelerate innovation. Data scientists and engineers use these tools to train and deploy models and manage ML operations.
How ITOps Use Machine Learning Today
While companies take advantage of ML to improve business opportunities and serve customers better through the scenarios explained in the previous section, ML can be especially helpful to ITOps organizations, no matter the industry. ITOps professionals know that when an incident occurs, and a network or service is impacted, it’s usually the result of a sequence of incidents. Whether it be a security breach or a complete system outage, an issue with one service can spread to affect another service. Before you know it, the problem is compromising overall availability, performance, and customer experience.
One obvious IT scenario where ML can help is the simple monitoring for and detection of network system failure. We’ve all read (or experienced) how missed signals can impact a company’s stability and even the bottom line for weeks or months, from simple appliance failures to forgotten software updates. And the responsibility for these issues—large and small—remains with ITOps, a sometimes-undersupported organization in this age of “do more with less” budgets.
It’s one thing to avoid disaster (failure detection), but ITOps teams can use ML to streamline processes and optimize the use of IT infrastructure in dramatic ways. For organizations like Managed Service Providers (MSPs) that manage networking and Wi-Fi, cloud services, IP telephony, hardware, software, and more, a little help can have a big impact. ITOps teams can use ML to gain visibility across entire environments, take advantage of data collection sets, set activity thresholds, and automate and manage alerts.
As Gartner notes in its Machine Learning Playbook for Data and Analytics Professionals, “Technology is often an enabler of business operations, and ML should not be considered any different.” This is particularly true for ITOps organizations, but it’s important to acknowledge the complexity of such projects and understand the overlap and differentiators among the related data sciences. In our next chapter, we will clarify the distinction between Machine Learning, Artificial Intelligence, and Algorithmic Intelligence.