Lessons from MURCS: Exploring the Connection between Scrum and Machine Learning


Machine learning is a field that relies heavily on empirical research and experimentation. But did you understand that this same approach will also be seen in an easy experiment involving a small black creature (Spooky Spider) with two legs which I made a decision to call Murcs?

Murcs’s primary goal is to climb higher and must learn tips on how to achieve this through trial and error. This strategy of climbing relies on knowing the particular configuration through which to bend each leg as a way to successfully ascend. Through repeated attempts, Murcs eventually discovers the right configuration that permits it to climb higher, resulting in increased ’’intelligence’’ and adaptableness to latest challenges.

Spooky Spider a.k.a Murcs

There are finite, but quite a few ways through which Murcs can bend its legs as a way to climb higher, but there are much more ways resulting in Murcs falling. In spite of everything, Murcs is just a chunk of software that may’t possibly know anything beyond what its creators tell it. Unfortunately, its creators only taught him tips on how to bend its legs and gave it a hit metric, on this case, Murcs’s success metric is pretty easy, the upper you climb — the higher you’re doing. Poor Murcs now has to try one-by-one each possible configuration it might probably bend its legs and after every trial, evaluate itself using the success metric. Each discovery then is rewarded with a rating on a scale of 10 (where 0 is the worst and 10 is the very best). Luckily, Murcs has a excellent memory and it might probably remember which configurations of bending its legs yield higher scores. Now, it’s only a matter of time until Murcs figures out or ’learns’ one of the best ways of bending its legs to climb higher.

In the event you read MURCS’s name from right to left, it says scrum, and other than being an easy wordplay, I’ll use it to speak in regards to the similarities between the scrum framework and machine learning.

Each Scrum and machine learning share a cyclical problem-solving process as their primary technique of tackling complex challenges. Within the Scrum framework, this process is facilitated through an iterative approach, wherein a dedicated team consistently produces high-quality work in regular intervals, relatively than . Similarly, within the realm of machine learning, an identical cycle of constructing, training, and testing models is utilized to deal with problems in a step-by-step manner.

Large datasets which might be subject to alter over time are ceaselessly used to coach machine learning models. Machine learning engineers must due to this fact be versatile and agile of their approach to model creation and retraining. The scrum team’s natural habitat involves adjusting the environment and priorities based on specific experience and business logic. There are a precise variety of ceremonies available for this, all of which serve to ensure transparency, inspection, and adaptation. So, the degree to which the present work matches the precise sprint goal and definition of done is repeatedly assessed, and the outcomes are swiftly incorporated within the shortest period of time.

Scrum methodology embraces the ability of incremental progress and iterative feedback loops, which ultimately result in the delivery of bite-sized, end-to-end functionality. Despite the short timeframe of two–4 weeks, commonly known as a sprint, Scrum further breaks down tasks into even smaller units of user-stories, able to being accomplished even inside a single day. This refined approach promotes a streamlined workflow and enables the team to deal with delivering high-quality outcomes in a timely and efficient manner. Similarly, within the realm of machine learning, data preparation processes will be likened to a mosaic, with various smaller components akin to cleansing, data transformation, and data augmentation coming together to create a masterpiece. By breaking down these larger tasks into smaller ones, teams can maximize their productivity and achieve their goals with precision and style.

Within the worlds of each scrum and machine learning, the important thing to success lies in a relentless pursuit of innovation and improvement. Through the strategy of conducting experiments, testing hypotheses, and validating assumptions, teams are in a position to gain useful insights and learn from practical situations. In each sprint retrospective, teams embrace the challenges and lessons encountered along the best way, turning them into opportunities for growth and improvement. Moreover, cross-functional and self-managing teams be certain that all areas of experience are represented and that everyone seems to be in a position to contribute their unique perspectives and skills. This collaborative approach fosters creativity and innovation and ultimately results in more practical and efficient solutions.

It is obvious that each ML and the Scrum framework are grounded within the principles of empiricism. Moreover, we are able to draw a parallel to the sector of psychology, where behavior that’s followed by positive consequences is more more likely to be repeated, while behavior that’s followed by negative consequences is less more likely to achieve this. That is the elemental concept of operant conditioning, which was introduced by the American psychologist B.F. Skinner within the mid-Twentieth century. In machine learning, we observe an identical strategy of learning through rewards and punishments, which is known as reinforcement learning. For instance, MURCS could also be rewarded with 10 points within the case of “correct” behavior, and it uses this feedback to find out whether it is moving in the precise direction toward its goal. By configuring its actions based on this reward system, MURCS is in a position to reach its objective and move forward. This will also be used as an ideal illustration for the scrum framework, because learning from the experience referred to as empiricism is its fundamental tenet, together with other principles akin to Lean considering, reducing waste, and specializing in the essentials.


What are your thoughts on this topic?
Let us know in the comments below.


0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments

Share this article

Recent posts

Would love your thoughts, please comment.x