AWS Certified Machine Learning – SpecialtyModelingMedium
A financial institution is developing a machine learning model to predict stock market volatility. They have collected a large dataset of historical stock prices, trading volumes, and macroeconomic indicators. The data exhibits strong temporal dependencies, with current values heavily influenced by past values. The team is exploring various model architectures. Which type of neural network is inherently designed to handle sequential data and capture long-range dependencies effectively?
- ARecurrent Neural Network (RNN)
- BFeedforward Neural Network (FNN)
- CConvolutional Neural Network (CNN)
- DGenerative Adversarial Network (GAN)
Show answer & explanationAnswer & explanation
Correct answer: A. Recurrent Neural Network (RNN)
Recurrent Neural Networks (RNNs) are specifically designed to process sequential data. They have internal memory that allows them to use information from previous steps in the sequence to influence the processing of the current step, making them suitable for time-series forecasting and data with temporal dependencies.
Why the other options are wrong
- B. FNNs process each input independently without considering the order or relationships between sequential data points, making them unsuitable for data with strong temporal dependencies.
- C. CNNs are primarily used for spatial data like images, although they can be adapted for 1D sequences, they are not inherently designed for long-range temporal dependencies like RNNs.
- D. GANs are used for generating new data samples that resemble the training data and are not primarily designed for processing and predicting based on sequential dependencies in time-series data.
Recurrent Neural Network (RNN)
A type of neural network designed to process sequential data by maintaining an internal state (memory) that captures information from previous elements in the sequence.
- Suitable for time series, natural language, and speech.
- Captures temporal dependencies.
- Suffers from vanishing/exploding gradients in vanilla form, often mitigated by LSTMs/GRUs.
Memory trick: RNNs remember the past, like a flowing river, each drop influencing the next.