IBM’s fame has gone wild in diverse fields because of its supercomputer, Watson. The AI, which has proven resourceful in innovative revolutions has been commissioned with a new development to war against cybercrimes.
Developers have been active over the years to see that fraud and other security threats such as email fraud and ransomware are minimized on various digital media. In this regard, IBM has been fully fledged to see to the healthiness of computer operatives; she recently sets to commission a developed an Artificial Intelligence (AI) that can scan a whole digital system to discover the possible threats and have them fixed.
Watson is set to be subjected to a learning system about how to run diagnostic checks on the data collected from various activities happening a system’s database, where it would analyze, point a fault, and block or remove threats. It is a great tool to be used in Cloud Email Security against cases like phishing and spyware.
Outliers and abnormalities are the usual detections to obtain the data for evaluation as a means to combat cybercrimes, for most AIs. An IBM report discloses that there are more than 200,000 highlights of these notices that are considered for security check. IBM’s Watson is set to handle the large volume of queries. Not just that, it is also to learn to use context to discern a type of threat. This is to be operative rather than a random pick on data that contain mutual words or terms with the ones in threats, as used by the previous AI solutions. Watson is to, therefore, make use of ungrouped or unstructured data to determine what really happened and how to solve anything that poses threat to breach the security of a cyberspace. With this, IBM gains the impression to be the first to launch an AI that can sift through an ungrouped data, link the various possibilities and detect the actual problem.
Considering the task to be done, IBM is taking Watson through the feeding stage, where it has to see and learn the various terms and the conditions that define security and threats. What makes this a huge project is that the Watson would have to learn to identify ‘campaign’, ‘symptom’, ‘incident’, ‘target’, ‘resource’, ‘impact’, ‘indicator’, and other terminologies; meaning, it has to understand relationships of the contents present in an array of unstructured data. For instance, it has to be able to work on data, such that it can identify malware indicators from an external source that is targeting the system which it watches over.
IBM has so much information to feed Watson, which is thought to go at the pace of about 15,000 security logs input per month, for a start, from its research library. The AI has to be, for a start, manually taught the meaning of each of the codes on the documents and how they relay.
As noted by researchers and developers, Watson has a low probable decision-making technique as an AI, beyond what has been reported about any other known AI. It is reliable on given orders, making use of data and magnitude to rapidly process queries based on the advancement of its complex algorithms.