AWS Certified AI PractitionerAI/ML and Generative AI FundamentalsMedium
A research team is developing a generative AI model to create novel protein structures based on known amino acid sequences. They need a model architecture that excels at understanding and generating complex sequential data with long-range dependencies. Which of the following generative AI model types would be most suitable for this task?
- ARestricted Boltzmann Machine (RBM)
- BVariational Autoencoder (VAE)
- CTransformer
- DGenerative Adversarial Network (GAN)
Show answer & explanationAnswer & explanation
Correct answer: C. Transformer
Transformers, particularly their attention mechanism, are highly effective at capturing long-range dependencies in sequential data. This makes them ideal for tasks like generating protein structures from amino acid sequences, which inherently have complex, distant relationships between elements.
Why the other options are wrong
- A. RBMs are an older type of generative model, primarily used for collaborative filtering or feature learning, and are not well-suited for complex sequence generation with long dependencies.
- B. VAEs are good for generating data and learning latent representations but are not inherently optimized for complex sequential dependencies as effectively as Transformers.
- D. GANs are excellent for generating realistic data, especially images, but struggle with long-range sequential dependencies without significant modifications.
Transformer Architecture
A neural network architecture, primarily used in natural language processing and increasingly in other domains, that relies on a self-attention mechanism to weigh the importance of different parts of the input data.
- Excels at capturing long-range dependencies.
- Parallelizable training, unlike sequential models like RNNs.
- Foundation of many large language models (LLMs).
Memory trick: Each model has its own superpower for generation.