What is machine learning?
“A learning machine is any device whose actions are influenced by past experience.” - Nils John Nilsson
A machine learning system “learns” from existing data, either classifying the data itself (unsupervised learning), or learning to recognise classifications with pre-categorized data (supervised learning). The basic technique has been around for decades, however interest in the topic has mostly been academic until recently.
Why the sudden interest?
Machine learning is computationally expensive; with the continual increase in computing performance, it is only now becoming possible to use the techniques in a commercially viable way.
What’s so special about machine learning?
The goal is to turn data into information, and information into insight.” - Carly Fiorina
Here are some recent technological advances that use machine learning:
- Facebook automatically recognises individuals in images
- Autonomous vehicle technology, such as Tesla’s “autopilot” system
- Analysing patient x-rays and providing more accurate medical diagnostics
- Reducing the number of spam emails
- Producing more accurate web search results
- Prevention and detection of credit card fraud
These types of problems are more challenging to solve with classic computational techniques, because there are so many dimensions to the problems.
Do I need a lot of data, and what sort of data?
“More data beats clever algorithms, but better data beats more data.” - Peter Norvig
Here are some examples of data which might be suitable:
- Text, such as tweets, emails, or comment posts
- Images (and lots of them)
- Complex, multi-dimensional numeric data
- Audio files
- Anything that may have an affect on the outcome you wish to understand!
If a skilled individual can make accurate decisions, based on a set of data, then a computer should be able to match or better the accuracy of those decisions, based on the same set of data.
How can machine learning benefit me?
“I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if it were a nail.” - Abraham Maslow
Machine learning can be a bit like that. However, there are some tasks it excels at, some it will never be good at, and others where the technology is just not ready yet.
If you’d like to know more, feel free to contact me at my consultancy company.