SimCLR<T>
Models & Types · AiDotNet.SelfSupervisedLearning
SimCLR: A Simple Framework for Contrastive Learning of Visual Representations.
Inheritance: ModelBase<T, Tensor<T>, Tensor<T>> → SelfSupervisedLearningMethodBase<T> → SimCLR<T>
For Beginners
SimCLR is one of the most influential self-supervised learning methods. It learns representations by maximizing agreement between differently augmented views of the same image using a contrastive loss.
How It Works
How SimCLR works:
Key hyperparameters:
Reference: Chen et al., “A Simple Framework for Contrastive Learning of Visual Representations” (ICML 2020)
Best for: Simple setup, strong performance, research baselines.
Pros: Simple architecture, strong performance, well-understood.
Cons: Requires large batch sizes for best performance.
Constructors
| Constructor | Summary |
|---|---|
SimCLR(INeuralNetwork<T> encoder, IProjectorHead<T> projector, SelfSupervisedLearningConfig<T> config) |
Initializes a new instance of the SimCLR class. |
Properties
| Property | Summary |
|---|---|
SelfSupervisedLearningMethodCategory Category { get; } |
|
bool RequiresMemoryBank { get; } |
|
bool UsesMomentumEncoder { get; } |
|
string Name { get; } |
Methods
| Method | Summary |
|---|---|
SimCLR<T> Create(INeuralNetwork<T> encoder, int encoderOutputDim, int projectionDim, int hiddenDim) |
Creates a SimCLR instance with default configuration. |