Variance is closely related to overfitting. High variance is usually caused by training a powerful model but with limited data, in other words, training with different sets of data would result in a different model.
Bias is closely related to underfitting. High bias is often caused by using a less powerful model.
