Data Science and machine learning have generated a lot of discussion surrounding them in the past few years. Between them, they are poised to take over and transform the ways that we do many things in our everyday lives: how we drive our cars, manage our money, and make smart decisions. If we asked you to define them and how they work together however, could you?
Data Science is the field of gathering, storing, cleaning, predicting on, and visualizing data. It’s a conglomerate and mash up of sub-disciplines from computer science, mathematics, statistics, and business. Machine learning is the practice of making models and algorithms, frequently referred to as “machines” by practitioners, to learn and predict based on new information. While we as humans have a tendency to want to delineate between things in our brain, the relationship between these two is muddy; most “data science” algorithms are in effect, machine learning algorithms. While data science also encompasses the fields of data mining and data engineering, machine learning is largely becoming the crux of what data scientists do.
All Machine Learning is AI, but not all AI is Machine Learning
When you hear someone discussing machine learning, they are speaking about artificial intelligence. As one of the core branches of AI, machine learning has brought us some of the most noteworthy advances towards “true-AI” – Siri on Apple’s iPhone, Amazon Alexa, and even the underbelly of your Facebook newsfeed. AI also contains many other fields such as computer vision and robotics, but we won’t be focusing on those.
Lately, there has been a lot of research into advancing a model called the artificial neural network. These models were originally developed to mimic the computational structure of the brain but recently have moved away from their biological inspirations towards new, mathematically inspired frontiers. Common forms of these networks are convolutional neural networks, which are frequently used for image processing, as well as the multipurpose recurrent neural networks, restricted Boltzmann machines, and autoencoders.
Within the study of artificial neural networks, there is a specialty area that you may have heard called deep learning. More than just an ominous phrase, deep learning is the use and implementation of deeply structured artificial neural networks. These networks have several layers of computation and are often constructed for tasks such as satellite image recognition, medical diagnosis, and speech recognition.
Machine Learning in Action
When we talk about machine learning models, we are talking about algorithms that make a prediction off the available data and learn based on new data. Chief among machine learning algorithms, you’ll see:
- Artificial Neural Networks
- Random Forests
- Bagging and Boosting Methods
- Clustering (K-Means, Hierarchical, DBSCAN)
- K-Nearest Neighbors
- Hidden Markov Models
Each time one model sees an error, it fixes this error by adjusting it’s parameters until it sees a new error. Such is the learning of machine learning!
On the federal side, many government agencies have routinely written on automating many processes of the federal government with machine learning, from finding duplicate transactional entries to recommending the best options for something looking to immigrate. For instance, a neural network could be used to actively scan the transactions of an agency looking for transactional fraud or unusual payments. Every time a new payment would run through the system, the algorithm would continually learn what a “proper” transaction looks like and would flag a transaction if it fell outside the usual, expected behavior based on it’s attributes.
From another angle, machine learning is being used across agencies to help our government interface with citizens every day. Numerous agencies that deal with citizens on a regular basis, from the IRS to the VA, are utilizing machine learning techniques such as recommendation engines to help those citizens find the easiest and most efficient way to pay their taxes or receive their benefits.
Machine learning can help any entity improve their analyses and their processes; it’s a method that not only describes and understands data, like statistics, but predicts, learns, and reasons – allowing for robust and dynamic analyses. The only question remains – how many facets of your organization can machine learning help?