AIN 311 MACHINE LEARNING BLOG 7 — MODEL COMPARISON
This week, we will compare the models we have prepared: CNN1, CNN2, MLP, and KNN. As mentioned in our previous blogs, we have already achieved the target accuracy of 70% in each of our models. We all know how significant recycling is for our world. To help you better understand the situation, I would like to provide some examples related to this.
Recycling used paper products reduces air pollution by approximately 73% and water pollution by 35%.
On the other hand, recycling used steel results in about a 97% reduction in mining waste compared to using virgin materials.
Recycling significantly reduces the use of raw materials. The reduction in raw material consumption, in turn, minimizes the environmental impact we impose. It contributes to less disruption of the natural balance, helping to preserve the ecological equilibrium.
MODEL ACCURACY COMPARISON
CNN1 MODEL ACCURACY → 0.7417
CNN2 MODEL ACCURACY → 0.7875
MLP MODEL ACCURACY → 0.7208
KNN MODEL ACCURACY → 0.7333
In our first blog, we mentioned that Germany has the highest recycling rate globally, with 67.2%. As you can see, the three models we developed have achieved even better results than this benchmark.
By creating smart trash bins and utilizing artificial intelligence applications, we demonstrated that it is possible to separate waste much more efficiently than current methods. Our CNN model, especially, reached an incredible accuracy of 78%, indicating its suitability for this task. We believe that by increasing the amount of data provided to our model, it is possible to raise the accuracy from 78% to around 85–90%. This would mean a much higher rate of recycling.
Remember, as long as the world exists, so do we. No individual can contend with the power of nature. Therefore, until smart recycling bins become widespread, everyone should contribute to recycling as much as possible.
Looking forward to meeting you in our next blog.
