DTU BIO 6 - AI/ML

What breakthrough is DeepMind's AlphaFold known for?

Protein folding prediction.

Which data type often requires AI and big data techniques for analysis in proteomics?

Mass spectrometry data.

How do neural networks assist in predicting protein structures?

They recognize patterns in sequence data.

Why is big data essential for protein-protein interaction studies?

Because of the complex and large number of potential interactions.

Which database is commonly used for retrieving protein structures for AI modeling?

Protein Data Bank (PDB).

What is the main challenge in protein folding that AI aims to address?

Predicting the 3D spatial arrangement of amino acids from its sequence.

What feature do many deep learning models in protein science use for sequence pattern recognition?

Convolutional layers.

How do AI, Machine Learning (ML), and Deep Learning (DL) relate to each other?

ML is a subset of AI, and DL is a method within ML focused on neural networks with three or more layers.

What is the main advantage of using transfer learning in protein prediction tasks?

It allows for faster and often more accurate model training.

What is a potential pitfall when training AI models on biased protein databases?

The model may not recognize less common proteins.

What differentiates Supervised, Unsupervised, and Reinforcement Learning?

Supervised Learning uses labeled data, Unsupervised Learning does not, and Reinforcement Learning involves an agent learning by interacting with the environment and receiving feedback.

What best describes Neural Networks in Machine Learning?

A series of algorithms that attempt to recognize underlying relationships in data mimicking human brain processes.

What does overfitting in machine learning models refer to?

When a model is so intricately tuned to training data that it performs poorly on new, unseen data.

Why is data split into training, validation, and test sets in machine learning?

To allow for model training, hyperparameter tuning, and unbiased evaluation on unseen data.

What role does the 'attention' mechanism play in deep learning models like Transformers?

It allows the model to focus on specific parts of the input data, weighing the importance of different information.

How does the 'denoising diffusion' model operate in deep learning?

It introduces noise to data and reverses the process, reconstructing the original data without noise.