This work proposes a convolutional analysis sparse coding based formulation for energy disaggregation. The resulting technique is shift-invariant and hence can learn to represent different devices ...