The future of HR lies in Deep Learning which is steroid machine learning. It uses a technique that gives machines an improved ability to find, and amplify, even the smallest patterns. This technique is called a deep neural network: deep because it has many layers of simple computational nodes that work together to search for data and deliver a final result in the form of prediction.
Neural networks were vaguely inspired by the inner workings of the human brain. The nodes are like neurons and the network is like the brain itself. But Hinton published his breakthrough at a time when neural networks had gone out of style. No one really knew how to train them, so they were not giving good results. The technique took almost 30 years to recover. But suddenly, it emerged from the abyss.
One last thing we should know in this introduction: machine learning (and deep) comes in three packages: supervised, unsupervised and reinforced.
In supervised learning, the most frequent, the data is labeled to indicate to the machine exactly what patterns to look for. Think of it as something like a tracking dog that will chase the targets once you know the wrapper you’re looking for. That’s what you are doing when you press play on a Netflix program: you are telling the algorithm to find similar programs.
In unsupervised learning, the data has no tags. The machine only searches for any pattern it can find. This is like letting a person check tons of different objects and classify them into groups with similar wrappers. Unsupervised techniques are not as popular because they have less obvious applications but curiously, they have gained strength in cybersecurity.
Finally, we have reinforcement learning, the last frontier of machine learning. A reinforcement algorithm learns by trial and error to achieve a clear objective. He tries many different things and is rewarded or penalized depending on whether his behaviors help or prevent him from reaching his goal. This is like when a child behaves well with a praise and affection. Reinforcement learning is the basis of Google’s AlphaGo, the program that surpasses the best human players in the complex Go game.
Applied to Human Resources, although the growth potential is wide, the current use of Machine Learning is limited and presents a dilemma that must be resolved in the future, related to the ability of machines to discover talent in human beings, beyond their hard and verifiable competencies, such as level of education, etc.
Software intelligence is transforming human resources. At the moment it has its main focus on recruitment processes, which in most cases is a very expensive and inefficient process where our goal is to find the best candidates among thousands of them, although we can find multiple application examples.
A first example would be the development of technology that would allow people to create job descriptions that are gender-neutral to attract the best possible candidates, whether male or female. This would boost a group of job seekers and a more balanced population of employees.
A second example is the training recommendations that employees could receive. On many occasions these employees have many training options, but often they cannot find what is most relevant to them; Therefore, these algorithms present the internal and external courses that best suit the employee’s development objectives based on many variables, including the skills that the employee intends to develop and the courses taken by other employees with similar professional objectives.
A third example will be Sentiment Analysis, which is a form of NLP (Natural Language Processing) that analyzes the social conversations that are generated on the Internet to identify opinions and extract the emotions (positive, negative or neutral) that these implicitly carry. With the sentiment analysis it is determined:
-Who is the subject of the opinion.
-About what is being said.
-How is the opinion: positive, negative or neutral.
This tool can be applied to words and expressions, as well as phrases, paragraphs and documents that we find in social networks, blogs, forums or review pages. The sentiment analysis will determine the hidden connotation behind the information that is subjective.
There are different systems of sentiment analysis:
-Analysis of feeling by polarity: Opinions are classified as very positive, positive, neutral, negative or very negative. This type of analysis is very Simple with reviews made with scoring mechanisms from 1 to 5, where number 1 is very negative and 5 is very positive.
-Analysis of feeling by type of emotion: The analysis detects emotions and specific feelings: happiness, sadness, anger, frustration, etc. For this, there is usually a list of words and the feelings with which they are usually related.
-Sentiment analysis by intention: This system interprets the comments according to the intention behind: Is it a complaint? A question? A request?
A fourth example is the Employee Attrittion through which we can predict which employees will remain in the company and which will not be based on several parameters as shown in the following example-
These four cases are clear examples in which Machine Learning elevates the role of human resources from tactical processes to strategic processes. Smart software is enabling the mechanics of workforce management, such as creating job applications, recommending courses or predicting which employees are more likely to leave the company, giving the possibility to react in time and apply corrective policies for those deficiencies.
From the business point of view, machine learning technology is an opportunity to drive greater efficiency and better efficiency in decision making. This will help everyone to make better decisions and, equally important, will give Human Resources a strategic and valuable voice at the executive level.
Prof Raul Villamarin Rodriguez