Notes on ScoreGrad

  1. 1. Architecture
    1. 1.1. Conditioner
    2. 1.2. Score Function

ScoreGrad is a EBM make use of iterative conditional SDE sampling via diffusion to perform multi-variate probabilistic time series prediction. Basically it uses RNN/LSTM/GRU to encode past time series as a condition and sample a probability distribution of predicting time series based on this.

  • Paper link: https://arxiv.org/abs/2106.10121
  • Code release: https://github.com/yantijin/ScoreGradPred

Architecture

Conditioner

As a conditional generation task, it use an RNN/LSTM and extracts last hidden state as feature. For a time

Score Function