Machine Learning and It’s Applications for L&D Professionals

Machine learning in layman’s term is the ability given to a computer or machine to learn without being explicitly programmed to do so.

Now think about a company IT or BFSI spending millions on training and struggling to get the right L&D strategy, still not able to customize learning as per individual’s need and skill set. 

This is where machine learning comes in to help facilitate the change. So, let’s find out what this means for a L&D professional. Well the application of Machine Learning (ML) is immense in the field of L&D and in this write up I will try to unravel the mystery behind Machine Learning and see how this can be practically applied to different facets of L&D. 

Machine Learning uses data mining and is generally divided in to two types of learning 

·      Supervised learning– In this type of machine learning we already know the right answers and provide a structured data set to the computer, we teach the computer how to do something, so that it helps us do the job faster and is able to predict right outputs (regression).

In L&D parlance, supervised learning can be effectively used in a lot of situations where you will need continuous outputs, especially in situations where the learner needs a customized recommendations post learning. For L&D professionals this will help them understand accurately in a matter of milliseconds as to where the learner is lacking and what skills need to be worked on for the learner to perform the job better. All this can be done with a simple algorithm (we will discuss this algorithm in the later part of the write up) . The amount of money and time we will be able to save for the organization will be unfathomable. 

·      Unsupervised learning– Is my personal favorite and this is where the machine learns by itself, what we need to provide is unstructured and unfiltered data set (these doesn’t have to be numbers always) - it understands behaviour, segmentation and data relevance by itself.

The unsupervised machine learning will find many use cases especially in the field of L&D, right from defining the Learning Metrics, conducting an automated TNA to segmenting entire base of Employees on the bases of their learning needs and their current skill sets.

Let’s take the use case of segmenting employees in an IT company according to their learning needs and skill set. This is very difficult to comprehend given the diversity, volume and dynamism around it. Let’s see how can we apply an unsupervised machine learning algorithm to arrive at solution. This becomes a lot easier and far more precise.

To arrive at these we can use multiple input sets as specified earlier, these don’t have to be particularly numbers we can use multiple sources of information that have no co-relation and may be very abstract from the other - Demographics of employees, performance, their 360 degree feedbacks, scores of multiple diagnostics they have taken, their survey feedbacks and may be even the way they communicate within the organization.

Once the above information is fed into an unsupervised machine learning algorithm, it will then help us come up with different patterns and segmentation. These outputs ironically, are called as Hypothesis.

And the Machine learning will keep making these outputs till the time the hypothesis does not come close to the reality (L&D Professional’s expectation). 

This way , the IT organization will have not only saved a lot of time, effort and resources, but it will also have a very accurate strategy and vision related to L&D.

I have spoken a lot about Datasets & Algorithms, but where and how will we as L&D professionals get to write these algorithms? Well, this can be learnt by using already existing applications like Octave and Matlab. The reasons to use these applications are that they already have these algorithms built at the backend, we just have to specify the Experience, Task and Performance (This can be tricky and need some prior study on ML).

There are a number of courses available online that help you learn Matlab. I personally recommend to start with Matlab, as the GUI is easier to comprehend. 

Machine Learning is actually not something of the far-off future. It's already here, shaping and simplifying the way we learn.