In the context of security, A.I. brings huge potential and has the power to transform performance. However, it is not (yet) a replacement for human evaluation and existing processes. Even with the improvements being made, security professionals shouldn’t look to artificial intelligence as a full solution, rather it is an extension and enhancement to our current operations.
Better monitoring requires fewer false alarms
A.I. has already proven itself as an excellent technology for false alarm reduction, eliminating more than 85% of the false alarms generated by previous systems while also increasing the percentage of positively identified threats. Partnering only with the best internationally recognised names in machine learning and A.I.,
What is A.I.?
Any technique that enables computers to mimic human intelligence can be classified as “artificial intelligence”. Techniques employed to mimic human intelligence can include logical rules, if-then rules, decision trees and machine learning. The most effective modern systems use machine learning, and often include the sub-set of machine learning called deep learning. But what do these terms actually mean?
Artificial Intelligence: Any technique that enables a computer to mimic some aspect(s) of human intelligence. Includes the use of if-then rules, logic, decision trees and machine learning.
Machine learning: A subset of A.I. that uses statistical techniques that enables machines to execute tasks without explicit instruction, using patterns and inference instead. Machine learning enables a computer to improve at tasks with experience.
Deep learning: A subset of machine learning that utilises multi-layered neural networks, permitting software to train itself and improve its performance on tasks, like speech and image recognition, through exposure to vast amounts of data.
Machine learning is best suited to tasks that have a lot of data, where each instance of the task is similar to the others and where each task has a clearly defined objective input and output – as is the case with human detection on video.
How does a computer learn and improve?
With machine learning, a set of parameters is applied to a given input to produce a desired output. At each learning step, the error between what the model produces and what the correct answer is gets measured. The parameters are then updated with respect to this error so that on the next try, the error should be smaller. Over a multitude of iterations, a high degree of accuracy can be achieved.
Why isn’t everyone developing A.I., what are the limitations?
The single largest limiting factor in the use of A.I. is the amount of data required to get an accurate, useful system – it is very data hungry! Without this data to learn from, it performs extremely poorly. Additionally, as deep learning models get more complex, they often take longer to train, so it can take longer to see performance gains or ever more resources are required to achieve the same result.
This is a very important consideration when choosing an A.I. provider, to ensure you get the best system/result, you need to ensure your partner is adequately resourced!
Another limitation is that it is very difficult to know why a deep learning system has made the decision it made. This has slowed down their adoption in areas where interpretability is important, such as medicine and credit scoring.
What improvements are being made?
There is ongoing research in order to improve the efficiency and effectiveness of machine learning techniques and the field is constantly evolving. Areas of research include:
How to reduce the amount of data required to train a deep learning model and how to speed up the learning process. Additionally, better graphical processing units (GPUs) allow for constant improvements in the number of calculations that can be done in the same amount of time.