Data Science And Statistics Study Guide Rated A+

What are Bayesians? {Ans: People who believe it makes sense to assume a probability distribution that data is drawn from and then see how that matches the real behaviour. They believe that frequentists are special form of Bayesians who are dishonest in their assumptions.}Why do some people describe machine learning models as hypotheses? {Ans: Because the form of model chosen is an assumption about the dataset you're training on. For example, a linear model implies that your dataset is linear! But, more generally, you can think of a machine learning model as a learning algorithm that provides you with a hypothesis. It selects a hypothesis from all the potential hypothesis that most closely matches your data. It does this using the model parameters. The model's parameters and underlying behaviour form the hypothesis you're making about the dataset. Your hypothesis is accurate if your model is accurate.}What do (B, C, H, W) and (C, H, W) mean in computer vision algorithms? {Ans: They define different input types to a model. For example, in PyTorch: https://pytorch.org/vision/stable/models/generated/torchvision.models.detection.fasterrcnn_resnet50_fpn_v2.html#torchvision.models.detection.FasterRCNN_ResNet50_FPN_V2_Weights In the first case (B, C, H, W) stand for the following: B - Batch size, the number of images that you're sending