Hey @voxey
To understand these concepts in depth, I would suggest you to have look at the deep learning and image processing courses provided by mathworks
Still there is a brief overview of the concepts which you asked:
1) Relu : It's an activation function which is used to introduce non-linearity to the network and helps our network to learn non-linear decision boundaries better
2) Pooling : pooling is an operation used in CNNs. It is done to reduce the size of feature maps. Also it makes the network robust by introducing rotational/translational changes
3) Convolution : It is an operation in CNNs. It's main purpose to extract features from the image
4) Inception : Architecture used in GoogleNet. Refer this link
5) Dropout : It is a regularization technique used to prevent the neural net from overfitting
6) weight : It is a learnable parameter, network learns it over training to perform the task for which it's trained
7) Reducing training time : You can explore many options like using transfer learning,training on GPU, reducing number of epochs etc
