DL, NN, ML Flashcards
How does ML work with data?
ML uses large datasets to train algorithms and build models. Feature engineering and tuning optimize the model’s performance.
What is feature engineering in ML?
The process of transforming raw data into a format that can better support the predictive capabilities of a machine learning model.
What are the types of ML?
Supervised learning, unsupervised learning, semi-supervised learning, and reinforcement learning.
What is supervised learning?
Supervised learning uses labelled data to train algorithms to classify data or make predictions.
It includes classification (categorising data) and regression (understanding relationships between variables).
What are the advantages and disadvantages of Supervised Learning?
Advantages: High accuracy and reliability due to labelled data.
Disadvantages: Requires labelled data (expensive and effort-intensive), less adaptable to new data, data security concerns.
What is Unsupervised Learning?
Uses unlabelled data to discover patterns and structures, such as clustering data based on similarities.
What are the advantages and disadvantages of Unsupervised Learning?
Advantages: Requires less effort, and can be adapted for unlabeled datasets.
Disadvantages: May produce inaccurate results without human validation, computationally complex.
What is Semi-Supervised Learning?
Semi-supervised learning combines a small amount of labelled data with a large amount of unlabeled data during training, reducing data acquisition costs and improving accuracy.
What is Reinforcement Learning?
A training method based on rewarding desired behaviours and/or
punishing undesired ones.
What is clustering?
A data mining technique for grouping unlabeled data based on their similarities or
differences.
How does a neural network process data?
It divides the data into multiple parts for training, passing it through the input layer, then hidden layers and coming out at the output layer.
State the 4 critical hyperparameters for DL.
Activation functions, Loss functions, Optimization algorithms, Batch size
State the 2 main types of DL.
Recurrent Neural Networks (RNN) and Convolutional Neural Networks (CNN)
What are the advantages and disadvantages of Deep Learning?
Advantages: Eliminates the need for data labelling, high accuracy, and optimization capabilities.
Disadvantages: Requires large amounts of data and high computing power, suffers from the “black box” problem (lack of explainability).
What is RNN?
A neural network with a recurrent structure that takes its previous state as input and outputs it.
Conveys information from a previous time step to the current one, utilizing past data.
