How to manage a Deep Learning team to improving an algorithm in production?
First of all, we split the process into a different part as pre-processing, choose the right model, algorithm implementation and integration.
1. In our team, there is a couple of people has a deep knowledge of the data. These persons can set up the training set in accord with the Deep Learning research that has the goal to find the best model in the literature for our specific problem.
2. The Deep Learning research has the assignment to understand if someone has faced the same problem in some kind of academic research. This is the best starting point to try to solve a real work problem. Bring academic research in an enterprise solution.
3. After trying to reproduce the same result of the academic research we start the real implementation of the algorithm. This means, hyperparameters implementation to find the best fit of our model.
4. The last part is very important. Once we find the best solution in the lab. We have to release this code in production. This is a critical part because sometimes the results are good in the lab but the model does not work very well in the real world.
#deeplearningteam
✴️ @AI_Python_EN
❇️ @AI_Python
🗣 @AI_Python_Arxiv
First of all, we split the process into a different part as pre-processing, choose the right model, algorithm implementation and integration.
1. In our team, there is a couple of people has a deep knowledge of the data. These persons can set up the training set in accord with the Deep Learning research that has the goal to find the best model in the literature for our specific problem.
2. The Deep Learning research has the assignment to understand if someone has faced the same problem in some kind of academic research. This is the best starting point to try to solve a real work problem. Bring academic research in an enterprise solution.
3. After trying to reproduce the same result of the academic research we start the real implementation of the algorithm. This means, hyperparameters implementation to find the best fit of our model.
4. The last part is very important. Once we find the best solution in the lab. We have to release this code in production. This is a critical part because sometimes the results are good in the lab but the model does not work very well in the real world.
#deeplearningteam
✴️ @AI_Python_EN
❇️ @AI_Python
🗣 @AI_Python_Arxiv