As the Internet of Things (IoT) scales up, manual processes and analysis performed by human beings become unviable because of the increasing amount of data.
This has resulted in growing demand for greater automation and speed of response in order to head-off issues early enough to avert disaster. These issues range from security to software or firmware updates that contain previously undiscovered bugs.
The concept of network intelligence paired with advanced data analytics is a means by which IoT organisations can address their need for speed as Syed ‘Z’ Hosain, the chief technology officer and co-founder of Aeris, tells Robin Duke-Woolley, the chief executive of Beecham Research.
RD-W: We’re going to talk about the use of artificial intelligence (AI) for network operations. Before we start, there is a lot of confusion in the market about the differences between AI, machine learning (ML) and deep learning (DL). Can you elaborate on that?
ZH: The differences are subtle. AI is the most general term, combining large amounts of data with fast, iterative processing and intelligent algorithms to learn from patterns or features in data and then apply actions as part of an automation system. ML is a subset of AI focusing on a specific learning task. DL is then a sub-set of ML, for example by using neural networks to rapidly determine which data is most relevant to analyse instead of being instructed.
RD-W: What is network intelligence and how does that relate to AI and ML?
ZH: With network intelligence, we’re trying to achieve certain key objectives. First, how can we optimise resource utilisation associated with the network? It’s not just about systems, it’s also about the people involved. We want to make sure that manual approaches are not a key method that people use for managing their data.
The reason is that the number of devices and the number of data sources generating the input for all of these AI and ML systems will just keep increasing dramatically, to the tune of billions of IoT devices. There is no way human and manual approaches will scale to make that work and achieve the success needed. So, what are we trying to do? We want to understand what’s happening. We want to learn about what’s happening, and then we want to provide that information after we analyse it and make recommendations for action.
The ultimate goal of AI is basically to take the actions necessary because human beings simply will not have the time to respond quickly enough in a practical way. Clearly, network intelligence requires use of standard data analytics techniques, including machine learning for understanding what’s important, finding the patterns, and then using the AI methods to implement automated actions.
