Published:
Online Q-learning can experience instabilities during training. This is because by using experience sampled sequentially from the environment leads to highly correlated gradient steps. Deep Q-networks (DQN) made deep reinforcement learning a viable approach to complex sequential control problems. In this section, we introduce the vanilla DQN algorithm. Next sections will discuss various improvements that have been proposed in the literature.
Published:
Linear regression using PyTorch C++ frontend.
Published:
Create a simple neural network using PyTorch C++ frontend.
Published:
The auto keyword changed its semantics starting from the C++11 standard. In this notebook, we will review its new semantics and examine its new flavors.
Published:
This post introduces gradient boosting. A nice introduction can be found at A Gentle Introduction to the Gradient Boosting Algorithm for Machine Learning.
Published:
In this notebook, we go over Spark’s resilient distributed dataset or RDD. The official programming guide can be found here. RDDs form the backbone of Spark’s data structures. The DataSet and DataFrame are based on RDD.
Published:
PyTorch is one of the well established libraries for modeling deep neural networks. The exposed Python API is the most commonly used one. However, the library also exposes bindings for C++. In this series of notebooks, I will try to demonstrate how to use the latter. I will be following to a large extent the documentation for the C++ frontend.
Published:
Brief introduction to reinforcement learning.
Published:
Brief introduction to hidden Markov models.
Published:
Matrix approximation with SVD