Gain insights into advanced deep learning and NLP by building 12 real-world projects using tools like TensorFlow and scikit-learn. Enhance your portfolio with industry-relevant skills.

dockerfile.tar.gz

python3

spa-job

In this course, you'll not only learn advanced deep learning concepts, but you'll also practice building some advanced deep learning and Natural Language Processing (NLP) projects. By the end, you will be able to utilize deep learning algorithms that are used at large in industry.

 

This is a project-based course with 12 projects in total. This will get you used to building real-world applications that are being used in a wide range of industries. You will be exposed to the most common tools used for machine learning projects including: NumPy, Matplotlib, scikit-learn, Tensorflow, and more.

 

It’s recommended that you have a firm grasp in these topic areas: Python basics, Numpy and Pandas, and Artificial Neural Networks.

 

Once you’re finished, you will have the experience to start building your own amazing projects, and some great new additions to your portfolio.

Building Advanced Deep Learning and NLP Projects

# Problem One
***Change the hyperparameters, the number of layers in our classifier, and the number of layers up to which we want to freeze the weight update process and see what the accuracy of the model is. How does it change?***

We will only change the model architecture here but you can even change the `epochs`, the number of parameters to freeze for the training process, etc.

# The ResNet50 model's output is going to be connected to this classifier.

av1 = GlobalAveragePooling2D()(model.output)

fc1 = Dense(256, activation = 'relu')(av1)

d1 = Dropout(0.5)(fc1)

fc2 = Dense(128, activation = 'relu')(d1)

d2 = Dropout(0.5)(fc2)

fc3 = Dense(64, activation = 'relu')(d2)

d3 = Dropout(0.5)(fc3)

fc4 = Dense(10, activation = 'softmax')(d3)

**Explanation:**
* We just added some more `Dense()` layers in our classifier architecture. Obviously, it will increase the number of parameters to train and, eventually, the time taken to train the model but it can give better accuracy as compared to the architecture used in our previous lessons.

We get an accuracy rise of `5%` with this architecture. You can perform the changes per your choice and look to obtain more accuracy.

# Problem two
***We used the ResNet50 model here. Your task is to use the MobileNet and AlexNet pre-trained model and build a classifier on the same dataset. You may also use any other image classification dataset.***

Below is the code snippet of the `MobileNet` model.

In this lesson, we will discuss the solutions for the assignments in this chapter.

Welcome to the Course

Project: Build a COVID-19 Detection System Using X-Rays

Project: Building a Pokemon Classifier Using Transfer Learning

Project: Text Generation Using Markov Chains

Word Embedding: Two Mini Projects

Project: IMDB Reviews Sentiment Analysis

Project: Deciphering Text Using Character-Level RNNs

Project: Emoji Predictor Using Transfer Learning in NLP

Final Exam

Where to Go Next?

Assignment Solution

Problem One