How a Neural Network Works:
A neural network (#neuralnetwork) uses rules it “learns” from patterns in data to construct a hidden layer of logic. The hidden layer then processes inputs, classifying them based on the experience of the model. In this example, the neural network has been trained to distinguish between valid and fraudulent credit card purchases.
This is not your mom’s apple pie or the good old days of case-based reasoning or fuzzy logic. (Although, the latter is still one of my favorite terms to say. Try it: fuzzzzyyyy logic. Rolls off the tongue, right?)…But I digress…
And, now, we’re back.
To give you a quick refresher:
Case based reasoning represents knowledge as a database of past cases and their solutions. The system uses a six-step process to generate solutions to new problems encountered by the user.
We’re talking old school, folks…Think to yourself, frustrating FAQ pages, where you type a question into a search box, only to have follow on questions prompt you for further clarification and with each one, further frustration. Oh and BTW, the same FAQ pages which e-commerce sites laughably call ‘customer support’ –
“ And, I wonder why your ASCI customer service scores are soo low Mr. or Mrs. e-Retailer :),” says this blogger facetiously, to her audience .
And, we’re not talking about fuzzy logic either – Simply put, fuzzy logic is fun to say, yes, and technically is:
–> Rule-based technology with exceptions (see arrow 4)
–> Represents linguistic categories (for example, “warm”, “hot”) as ranges of values
–> Describes a particular phenomenon or process and then represents in a diminutive number of flexible rules
–> Provides solutions to scenarios typically difficult to represent with succinct IF-THEN rules
(Graphic: Take a thermostat in your home and assign membership functions for the input called temperature. This becomes part of the logic of the thermostat to control the room temperature. Membership functions translate linguistic expressions such as “warm” or “cool” into quantifiable numbers that computer systems can then consume and manipulate.)
Nope, we are talking Neural Networks – the absolute Bees-Knees in my mind, right up there with social intelligence and my family (in no specific order :):
–> Find patterns and relationships in massive amounts of data that are too complicated for human to analyze
–> “Learn” patterns by searching for relationships, building models, and correcting over and over again model’s own mistakes
–> Humans “train” network by feeding it training data for which inputs produce known set of outputs or conclusions, to help neural network learn correct solution by example
–> Neural network applications in medicine, science, and business address problems in pattern classification, prediction, financial analysis, and control and optimization
Remember folks: Knowledge is power and definitely an asset. Want to know more? I discuss this and other intangibles further in part 1 of a multi-part study I am conducting called:
Musings on ML, Deep Learning & AI 