Fundamentals of Neural Networks - Lab 2 - Sequence to Sequence Stock Candlestick Forecast

The video tutorial covers the basics of neural networks, focusing on feedforward and recurrent architectures, particularly LSTMs. It explains sequence prediction, backpropagation, and loss functions. The tutorial guides viewers through data preparation, splitting data into training and testing sets, and designing a model architecture using Keras. It also covers training the model, evaluating its performance, and tuning it for better results. The tutorial concludes with a demonstration of the model's predictions and discusses areas for improvement.

What is the primary function of a long short-term memory (LSTM) network in the context of sequence prediction?

• To store data permanently

• To predict a single numerical value

• To predict a sequence of values

• To perform image recognition

In backpropagation, what is the main purpose of the loss function?

• To generate random predictions

• To compare predicted values with actual values

• To eliminate the need for training data

• To increase the complexity of the model

What is the role of the explanatory data matrix (X) in the model?

• It is the final prediction of the model

• It is the output of the model

• It is used to calculate the loss function

• It is the data the model learns from

How does the model handle the difference in dimensions between X and Y matrices?

• By using a different loss function

• By reshaping the matrices to match

• By adding more layers to the model

• By ignoring the extra dimensions

Why is dropout used in the model architecture?

• To increase the number of neurons

• To prevent overfitting by randomly dropping neurons

• To speed up the training process

• To ensure all neurons are used equally

What is the significance of using a sequential model in Keras?

• It automatically optimizes the model

• It is the only model type available in Keras

• It allows for parallel processing of layers

• It ensures layers are added in a specific order

What is the purpose of using the Adam optimizer in model training?

• To reduce the number of training epochs

• To increase the size of the dataset

• To improve the accuracy of predictions

• To simplify the model architecture

What challenge is highlighted in the visualization of the model's predictions?

• The model predicts only upward trends

• The model's predictions are too volatile

• The model fails to predict any trends

• The model's predictions are too smooth

What is a potential solution to improve the model's prediction accuracy?

• Reducing the size of the dataset

• Switching to a different programming language

• Increasing the number of training epochs

• Using fewer layers in the model

What does the term 'inside month' refer to in stock prediction?

• A month with the lowest stock prices

• A month with the highest stock prices

• A month where prices stay within the previous month's range

• A month with no stock price changes