The general idea behind artificial intelligence (A.I.) is giving machines the ability to think like human beings. This has been a mysterious and sexy topic in the media and pop culture for decades.
However, we are a long ways off from machines being able to think and act independently. To some this is unfortunate, and to others comforting. YAI does shape our modern world in profound ways that most people do not even realize.
There is a sub-field within A.I. called machine learning. Machine learning is a branch of artificial intelligence that allows software to become more accurate at predicting specific outcomes, without explicit programming to do so. Essentially, you are “training” the machine to solve a problem.
As of recently, machine learning has begun to impact our lives in many ways. Computers finally have the speed and ability to keep up with machine learning tasks. For example, it is the concept of self-driving cars, speech recognition, and even online shopping.
What’s the difference?
There are two types of machine learning to be aware of. The first is supervised machine learning, and the second is unsupervised machine learning.
Supervised learning allows a computer to “learn” about a subject using “labeled data.” Labeled data tells us about an outcome that is already known. The goal is to give a system a sample of data for it to be able to predict outcomes accurately. The more data provided, the more accurate the predictions will be. A good example of how this can be applied is in the medical field, for instance, in cancer research.
Say we are looking at human cells, and we know whether a cell is cancerous or not. That is the label. We also have lots of other data about the given cells, for example, chemical composition, size, and many other things. These things that we know about the subject are “features.”
The goal is that using these features and labels, with machine learning, we can look at a whole new cell in the lab and be able to predict whether or not it is cancerous. The concept and math behind supervised learning are complex. A simpler way to think about it is whether or not we can get a computer to accurately predict something given things we already know about that subject.
By contrast, unsupervised machine learning does not have labeled data. Looking at our medical example, say we have collected a lot of information about a lot of cells. Using machine learning, we would try to find out several things: Are there common trends in the data? Which cells, if any, share common characteristics?
The machine does not initially know which cells are cancerous or not, but using this data, it can categorize them based on their other shared features. Machine learning compares thousands of these “features” simultaneously.
How does CUJO AI use machine learning?
So, when we say CUJO uses AI, that means an application of machine learning to protect your home network. CUJO AI uses a Supervised machine learning algorithm. Using labeled data, CUJO AI makes predictions about the behavior on your home network.
Specifically, we use this technique to monitor whether your devices are behaving correctly. It allows us to determine if you are browsing sites that house potential phishing, malware, or other types of malicious content. CUJO AI can do this without invading user privacy. Using machine learning, we can determine if your encrypted connection is secure.
The benefit of this type of technique that CUJO AI employs is that we do not need to know about a particular threat before it happens to defend from it. The term “zero-day” attack refers to a threat that has never been used before in a public forum.
Due to it being so new, traditional methods of antivirus cannot defend against it. This is where CUJO AI comes is. We don’t need to know about a particular attack before it happens. CUJO AI can predict malicious behavior and stop it in its tracks.
CUJO AI is always getting smarter
CUJO AI uses machine learning to stay ahead of hackers. Hackers are constantly finding new, complicated, and more sophisticated ways to infiltrate networks. But we get smarter too.
CUJO AI protects a growing number of global home networks, and we can collect some metadata about traffic on your network, without invading our your privacy. CUJO AI then can combine all of these data snippets to come up with more accurate predictions, via machine learning.
This process is regularly occurring. As we observe a new behavior on a customer’s network, we classify it and determine whether it is malicious, effectively making the entire CUJO AI network smarter. Everyone benefits from this process of learning, except criminal hackers.