Pytorch vs MXNet. Which is faster?
I’ve written about GluonTs before and it’s a great library for time series forecasting, however did you know that most of it’s models are built on top of MXNet and for some architectures there is a...
Neural Aspect
The F1 Score Time Series Model Championships
With the British Grand Prix just around the corner it’s a GO, GO, GO as we launch our Formula 1 themed Time Series Model Championships. Yes, that’s right folks, us at the totally not made up Federa...
I don't care what GenAI is and you shouldn't either
When did you first hear the term GenAI or GenerativeAI? A year ago or 2 years maybe. Now check out this video generated with the help from an AI model called DeepDream by Google. DeepDream Groce...
State of the art timeseries forecasting with 1Cycle scheduling
1. Introduction In this, a follow up to my two previous posts DeepAR and superconvergence, I will show you how to obtain state of the art performance using my simplified version of the DeepAR mode...
Supercharge your Neural Networks with Super-convergence
1. Introduction Want to know how to get a performance boost from your neural network for virtually no extra effort? Well I’m going to show you how to do it with a learning rate scheduling system c...
A fistful of MASE: Deconstructing DeepAR
1. Introduction Hey Amigos! Four years since its release, DeepAR is an old timer in the world of neural networks for time series forecasting, but it can still outshoot many of the young guns out t...
Should we predict covariates in an LSTM for timeseries forecasting?
1. Introduction This blog accompanies a paper I wrote on the effect of covariates on LSTM networks for timeseries forecasting, which you can find on arxiv. In business, leading indicators serve a...
A look at Teacher Forcing and it's alternatives
1. Introduction In sequence modelling we often use a model that is auto-regressive. That is to say that the output of one time step serves as the input into the next time step. This concept us used...
Human-level control through deep reinforcement learning
Playing Atari with Deep Reinforcement Learning: A Journey into the World of DQN Part 2 This is the second in a series posts where we revisit the landmark papers from DeepMind on Deep Q Learning. I...
Revisiting Playing Atari with Deep Reinforcement Learning
Playing Atari with Deep Reinforcement Learning: A Journey into the World of DQN It’s been over a decade since Mnih et al. published their groundbreaking paper, marking the first successful fusion ...