State the necessity of exponentially moving weighted average. In batch norm, the mean and variance at test time are calculated using the means and variances of various mini batches at training time using exponentiality moving weighted average. The means across four minibatches at training time are 0.78, 0.92, 0.82 and 0.29 respectively. The variances across the same four minibatches are 0.14,0.12, 0.1, and 0.37 respectively. Compute the mean and variance at test time with and without bias correction. What conclusions can you draw based on the obtained results? Momentum hyperparameter is 0.9