Starting my journey as the AI/ML engineer from Android Engineer... My current goal is to win a hackathon/competition related to this domain within 60 days and secure a job/contract related within 100 days.
I will be documenting the concepts learned and things experimented in this series of dev.to posts.
My Curiosity
First of all, I started with the following curiosity in the mind:
- What's the role of machine learning in software engineering?
- What problem will it try to solve, or solve which traditional approach of programming cannot?
- What are its usecases?
My Discovery
Till now, what I discovered is that machine learning can be a replacement/alternative/substitute of conditional statements (if..else, while loop, etc.) of the conventional programming approach.
When you have just a few branches, conditional statements work perfectly. But when the number of rules becomes huge — like trying to decide something based on hundreds or thousands of possible conditions — hand-coding all those if-else paths turns into a tangled mess that’s hard to maintain and scale.
Here to solve this we train the model on the data (condition, value). We predict the value from the condition using this trained model.
The Sweet Spot (Still Searching)
However, I have not discovered the sweet spot for machine learning from which branching of conditional statements are to be replaced. But I am pretty sure that machine learning should not be imposed on the place where conditional statement branches will do the thing.
Is my understanding correct? Please give your feedback and view.
This much for today from my side. Will keep updating about my journey.
Top comments (1)
your intuition about keeping simple rules in code is sound. i would add that machine learning is not a direct replacement for every branch. it is useful when the rules are hard to write, the input is uncertain, and you have examples with a clear outcome. for the next post, try one small classifier and compare it with a rule based baseline using the same test set. that will show the tradeoff in a clear way.