Machine-Learning Part II: How the Terminator Uses Deep Learning
Welcome back to Part II of our blog series explaining how the artificial intelligence evolution in the Terminator movies parallels how our data is used in real life. Continuing from last week’s article, we’re going to talk about how data is processed.
Once the collected data from all of your internet-enabled devices is gathered, it is sent to the next phase in the process called machine-learning. Machine-learning is the science of getting computers to act without being explicitly programmed. It leverages the use of algorithms which can learn from and make predictions on data, and then that data is used to make data-driven predictions, or “smart” decisions.
Machine-learning is the first step of intelligent automation technology. It helps complete repeatable, time-consuming tasks for accounts receivable teams. When you combine the efficiency and accuracy of machine-learning technology with the “educated” manual input, you gain a powerful accounts receivable (AR) assisting tool. Once machine-learning collects the data from these devices, it begins working on completing tasks, but it still needs guidance on how to prioritize each job.
Manual input teaches technology how to complete tasks moving forward, preventing the need for continuous human input. Machine-learning is malleable as it lacks a concrete structure, and becomes more organized and functional as you input manual processes.
According to Billtrust’s Director of Product Strategy, Alex Ross, “Machine-learning has the ability to learn from human inputs and behavior to create an automated process. Doing this in the fintech space allows companies to leave those mundane tasks to a computer while people can spend time on more critical thinking projects.”
Ross explains how machine-learning technology is used in real-world applications. For example, accounts receivable departments can use machine-learning processing within cash application software. “By using machine-learning to apply cash to 80% of the payments received, it frees up time for people to focus on those 20% of exceptions that otherwise could be rushed due to time constraints. Also, not forcing people to do mundane and repetitive tasks can lead to higher morale and less turnover.”
After machine-learning, the data goes through an education stage or filter which allows the next phase of processing for the incoming data, without requiring human input. But where does the education come from? It comes from your operating system, your ERP, or your database software, all of which act as the brain behind this education stage. Once the data makes its way through the education filter, it goes to deep learning technology.
Similar to machine-learning, deep learning can complete repeatable tasks with ease, however, the education filter has provided the input needed to be able to organize all the data into silos. Deep learning is the application of artificial neural networks (ANNs) to learning tasks that contain more than one hidden layer. Deep learning is part of a broader family of machine-learning methods based on learning data representations, as opposed to task-specific algorithms. Learning can be supervised, partially supervised or unsupervised. For accounts receivable (AR) teams, this allows systems to complete repetitive processing accurately, freeing up employees to handle more strategic tasks.
The silos build in structure and routines so the technology knows how to complete the tasks, and in which order they should be completed. The data is more neatly organized, however, there still data that didn’t receive input from the education filter. This data can further be managed in the next step of the process, which we will learn about in the next blog post.
To get back to my favorite movie, you can also think about machine-learning is like the upgraded T-1000 Terminator model. The T-1000 was malleable and could adapt to the data it was processing and receiving. The technology would know how to complete the task and can adapt to constraints on the fly. Similar to how the T-1000 can shapeshift to represent different forms, such as a police officer, to manipulate humans to help it achieve its goal of destruction, or to simply process an incoming payment without wasting time.
Deep learning technology is more structured, thanks to the education filter, and now can complete tasks like machine-learning, but with the ability to prioritize tasks. This technology is similar to the T-H Terminator model in which the physical structure of the machine appeared to be a human. This Terminator represents the ability for the Terminators and Skynet to learn from its past mistakes to take on a more evolved form.
Stay tuned for Part III of our Terminator blog series, when we get to look for Sarah Connor.
Looking for the start of this series? Click here for Part 1, (Terminator meets IoT)