Automated Machine Learning

Why is Automated Machine Learning Important?

Manually constructing a machine learning model is a multistep process that requires domain knowledge, mathematical expertise, and computer science skills – which is a lot to ask of one company, let alone one data scientist. Not only that, there are countless opportunities for human error and bias, which degrades model accuracy and devalues the insights you might get from the model. Automated machine learning enables organizations to use the baked-in knowledge of data scientists without expending time and money to develop the capabilities themselves, simultaneously improving return on investment in data science initiatives and reducing the amount of time it takes to capture value.

Automated machine learning makes it possible for businesses in every industry – health care sector, marketing, retail, sports, manufacturing, and more – to leverage machine learning and AI technology — technology previously only available to organizations with vast resources at their disposal. By automating most of the modeling tasks necessary in order to develop and deploy machine learning models, automated machine learning enables business users to implement machine learning solutions with ease, thereby allowing an organization’s data scientists to focus on more complex problems.

HYPER PARAMETERS:What is a hyperparameter?

A hyperparameter is a parameter that is set before the learning process begins. These parameters are tunable and can directly affect how well a model trains. Some examples of hyperparameters in machine learning:

1.Learning Rate

2.Number of Epochs

3.Momentum

4.Regularization constant

5.Number of branches in a decision tree

6.Number of clusters in a clustering algorithm (like k-means)

Optimizing Hyperparameters

Hyperparameters can have a direct impact on the training of machine learning algorithms. Thus, in order to achieve maximal performance, it is important to understand how to optimize them. Here are some common strategies for optimizing hyperparameters:

Grid Search: Search a set of manually predefined hyperparameters for the best performing hyperparameter. Use that value. (This is the traditional method)

Random Search: Similar to grid search, but replaces the exhaustive search with random search. This can outperform grid search when only a small number of hyperparameters are needed to actually optimize the algorithm.

Bayesian Optimization: Builds a probabilistic model of the function mapping from hyperparameter values to the target evaluated on a validation set.

Gradient-Based Optimization: Compute gradient using hyperparameters and then optimize hyperparameters using gradient descent.

Evolutionary Optimization

: Uses evolutionary algorithms (e.g. genetic functions) to search the space of possible hyperparameters.

Task description

1. Create container image that’s has Python3 and Keras or numpy installed using dockerfile 

 syntax:
    docker build -t <imagename> <path to Dockerfile>
      

    docker build -t sklearn_ml .     

(here "." means the current location/folder)

                                           docker build -t NN .

Job1 : Pull the Github repo automatically when some developers push repo to Github.

2. When we launch this image, it should automatically starts train the model in the container.

By looking at the code or program file, Jenkins should automatically start the respective machine learning software installed interpreter install image container to deploy code and start training( eg. If code uses CNN, then Jenkins should start the container that has already installed all the softwares required for the cnn processing).

I have created a python script which would differentiate among the codes and also copies the data as well as the code file in their respective folders which will be mounted later on to the container working directory .

Job3 : Train your model and predict accuracy or metrics.

Job4 : if metrics accuracy is less than 80% , then tweak the machine learning model architecture.

I have created a python script which would first train as it is provided by the developer .

Then it do some auto feature scaling and addition of layers to the Neural Network Architecture and every time it checks for the accuracy threshold and if it crosses the threshold the developer is informed through mail.

AUTO FEATURE SCALING

Auto addition of Dense layers in case of ANN and CNN

I have set a limit of 3 -4 layers of Dense if the accuracy doesnot increases it changes the units ie number of neurons in each layer

AUTO ADDITION OF CONVOLUTION LAYER

WITH DIFFERENT NUMBER OF FILTERS,FILTER_SIZE

Job5: Retrain the model or notify that the best model is being created

HERE add means to add the number of layers.

Tweaking by adding layers ,changing numbers of neurons,

Here we got the best accuracy in 4TH job of retraining

then the mail is sent to the developer.

For this job i have used mailx

Mail code

Create One extra job: for monitor : If container where app is running. fails due to any reason then this job should automatically start the container again from where the last trained model left

Thanks "Vimal Sir " for guiding and leading us towards the world of creators not the users by providing us the #rightEducation.

Github repo login:

https://github.com/raghav1674/MLOPS/blob/master/README.md