What is the key difference between visualization techniques using t$-$SNE $($t$-$distributed Stochastic Neighbor Embedding$)$ and PCA $($Principal Component Analysis$)?$

Group of answer choices

t$-$SNE produces interpretable components that directly represent the original features, while PCA produces latent variables that are linear combinations of the original features.

PCA is computationally more efficient than t$-$SNE, making it preferable for large datasets.

t$-$SNE preserves local structure in the data, making it suitable for revealing clusters and non$-$linear relationships, while PCA emphasizes global structure and linear relationships.

t$-$SNE is a linear dimensionality reduction technique, while PCA is a non$-$linear technique.