6th Grade Convolutional Neural Network Lesson Plan

Topic: Deep Learning for Text Classification

Objectives & Outcomes

• Given a set of documents, learn to classify each document as either "negative" or "positive" with respect to a particular topic (e.g. "cancer").

Materials

• A set of documents that have already been labeled as positive or negative with respect to the topic of interest
• A deep learning framework (e.g. TensorFlow)
• A large collection of text data (e.g. Wikipedia)
• A computer with enough processing power to run the deep learning model

Step 1: Prepare the Data

• Collect a set of documents that have already been labeled as positive or negative with respect to the topic of interest. For example, if the topic is "cancer", collect a set of documents that have already been labeled as either "cancer is a serious disease" or "cancer is not a serious disease".
• Prepare the documents into a tabular matrix where the rows represent documents and the columns represent the words in the documents. This is commonly referred to as a "document-word matrix". For example, if the document-word matrix has

elements, then the ith row will give the number of times the word appears in the ith document. For example, if the documents are reviews about different restaurants, then the ith row will give the number of times the word "cheese" appears in the ith review.

• Label each document as either "positive" or "negative" with respect to the topic of interest.

Step 2: Prepare the Training Data

• Collect a set of documents that have not already been labeled as positive or negative with respect to the topic of interest. For example, if the topic is "cancer", collect a set of documents that have not already been labeled as either "cancer is a serious disease" or "cancer is not a serious disease". This set will be used to train the deep learning model.
• Label each document as either "positive" or "negative" with respect to the topic of interest.

Step 3: Build the Model

• Build a convolutional neural network for text classification using the large collection of text data (e.g. Wikipedia).
• Train the model on the prepared training set.

Step 4: Evaluate the Model

• Use the model to classify the documents in the set that has not already been labeled as positive or negative with respect to the topic of interest.
• Compute the accuracy of the model on the test set.

Guided Practice

• Ask students to work in small groups to implement a convolutional neural network for text classification.
• Have each group present their model to the class and discuss the decisions they made during implementation.
• Ask students to identify the inputs, outputs, and learnable parameters of their model.
• Discuss the advantage of using a convolutional neural network over other text classification methods.

Independent Practice

• Ask students to choose a topic and create a set of training data for their convolutional neural network.
• Have students train and test their model on the training data.
• Ask students to present their model to the class and discuss the decisions they made during training.
• Ask students to create a test set of data and evaluate the performance of their model on the test set.

Closure

• Ask students to share what they learned about convolutional neural networks and text classification.
• Ask students to reflect on the process of creating their own model and discuss any challenges they faced.

Assessment

• Observe students during the guided and independent practice activities to assess their understanding of the subject matter and their ability to apply it in creating a model.
• Collect and review their models to assess their ability to accurately classify the texts.