Cognitive Systems Like Watson Are the Future of Network Management
Humanity’s thirst for technology is straining the world’s network infrastructure. The things people want, from cat videos to connected cars, require significant bandwidth, ubiquitous connectivity and endless storage and compute capacity.
People don’t just want the things technology can provide; they want them instantly, with no inconvenience. As a result, human demand for services — and human determination to have them right this moment — eclipses the human network administrator’s ability to deliver those services.
Cognitive systems can learn and make better predictions than humans. They can make faster decisions and execute them more quickly. They’re the logical choices to manage emerging network technologies such as software-defined networking (SDN). They’re not only an option; they’re inevitable.
Winning at More Than ‘Jeopardy’
The most famous of the world’s cognitive systems is IBM’s Watson. The computer is famous for processing natural language with incredible speed and flexibility — as well as its definitive “Jeopardy” prowess.
Watson’s language processing capability is made possible by its ability to think more like a human. It relies on human inputs of knowledge and human interaction to learn, but it delivers evidence-based recommendations and acts more quickly than any human ever could.
Humans upload a body of knowledge into Watson, curating the content to ensure no questionable sources enter its knowledge base. Watson reviews the information, processing not only the language and jargon, but also the mode of thinking within a field.
Watson organizes the information given to it, using indices and metadata to structure the knowledge. It creates a knowledge graph that forms connections between data, giving it the ability to answer questions more precisely.
Humans upload pairs of questions and answers relevant to the field for Watson to analyze. But Watson does more than memorize these questions and answers; it also learns to uncover patterns that better help it answer similar questions. Ongoing interaction with humans enables it to fine-tune its processing, and it receives constant updates to its corpus of knowledge. Eventually, Watson answers questions, offering insights humans might miss and delivering recommendations based on evidence-based statistical analysis.
Watson processes information like a human, but it operates with the speed of a supercomputer. It also never stops learning, which means it makes better and better decisions based on experience and human feedback.
Cognitive Systems and Tomorrow’s Network
SDN gives administrators the power to centralize the network’s control plane instead of configuring each individual appliance. They can rearrange the plumbing at the touch of a button, making centralized adjustments to the network’s virtualized appliances.
Automating tasks like failure management and scaling means significantly better network performance. Still, learning and improvements happen at a human pace. Cognitive systems centered around computers like Watson will change that.
In addition to delivering better network performance, cognitive systems can improve security because they can make sense of the tsunami of log data and alerts from the network. They can correlate those alerts with CVE information and a general understanding of the current threat environment, and deliver automated responses to potential attacks.
Balancing Risk and Reward
Watson is already helping doctors at Memorial Sloan Kettering prolong the lives of cancer patients. Imagine how many lives could be saved by a cognitive network capable of scaling instantly and rerouting telecommunications during a natural disaster.
Even as attack scenarios feed people’s worst fears about artificial intelligence, human demand for better and faster services remains the ultimate driver toward cognitive systems development. The potential rewards of AI-managed networks are simply too compelling to ignore.
Image Source: Wikimedia Commons