316 - Optimizing Steel Strength using Metaheuristic algorithms (e.g., Genetic)

Code generated in the video can be downloaded from here:
github.com/bns...
In this example, we will work with the steel alloy data set.​
Download from here: www.kaggle.com...
The data set contains the elemental composition of different alloys and their respective yield and tensile strengths. ​
A machine learning model can be trained on this data, allowing us to predict the strength of an alloy based on its chemical composition. ​
But, for this exercise, let us try to find the optimized alloy composition with the best yield strength.​
Let us explore metaheuristic approaches, especially the genetic algorithm and the differential evolution algorithm.​
Note: Differential evolution (DE) is quite similar to the genetic algorithm (GA) with a few differences. DE relies on the distance and directional information through unit vectors for reproduction. Also, in DE, the crossover is applied after mutation unlike GA. In addition, the mutation operator is not created from a probability distribution, but from the creation of the unit vector.​

[@timi_t_codes]

Thanks for this. looking forward to the next video 🔥

[@antaripgiri142]

Hi @DigitalSreeni, saw all the videos on genetic algorithms, I absolutely like the way you explain. Wanted to know if there is a upcoming video on Reduced Order Modelling (ROMs) which essentially combines physics or first principle with Machine Learning - would like to see your explanation and take on it. Its a topic I quite find interesting but conceptually not yet able to grasp it. Just a request :).

[@VedikaaDhiman]

How can we perform this code with ANN model?

[@abdulsami5843]

how do we adjust for constraints ?

[@omeralmani9755]

thanks for the video. I am just rying to understand the true power of GA. I am wondering why can't we simply find the minimum yeild strength from CSV opened in excel using MIN commmand. The elements corresponding to that MIN strength would be similar to what we found using GA. The reason is GA uses training data and that training data must cover a wide parameter space otherwise the objective function from random forest won't be accurate? So what's really the point of GA when the parameter space has been already evaluated? Could you please explain? Thanks

[@陽明交大-高明秀]

Great! can we you Q-learnig or PPO to optimize the trength?

[@xuchengtong2013]

Thanks🥰 Hope there could be a theoretical explanation of evolutionary history from YOLOv1 to v9 and their applications on microscope image multiclasses object segmentation!

[@DigitalSreeni]

Please refer to my YOLO video. kzbin.info/www/bejne/gILCg4WZfbF-fMU

[@RAZZKIRAN]

please provide colab note book also

[@DigitalSreeni]

Please look at the video description for the link to code.