Inference for Sequential Sampling Models
I link to a few more recent projects of mine, which were part of my PhD research.
Rapid Simulation of Sequential Sampling Models (ssms)
This is a lightweight package to forward simulate a variety of Sequential Sampling Models of interest
in Cognitive Science. To computational efficiency, the simulators are written in Cython.
The package integrates nicely with the LANFactory package below, for which we ssms can provide training data in the
expected format.
To install ssms type the following into your console:
pip install git+https://github.com/AlexanderFengler/ssm_simulators@main
Train Likelihood Approximation Networks (LANFactory)
Train Likelihood Approximation Networks (LANs) with this leight weight package.
The package asks for training data in a specific format which the ssms package above provides.
You can of course provide your own training data instead.
To install lanfactory type the following into your console:
pip install git+https://github.com/AlexanderFengler/LANfactory@main
HDDM
HDDM is a python toolbox for Bayesian hierarchical inference of cognitive process models.
Originally developed with specific focus on the Diffusion Decision Model,
it has since been extended to incorporate a whole host of other Sequential Sampling Models. These extensions make use of
Likelihood Approximation Networks.
I was mainly responsible for these extensions and continue to work on maintaining the project.
To install HDDM type the following into your console:
pip install git+https://github.com/hddm-devs/hddm
Tutorials
The HDDM package comes with various tutorials, especially concerning the recent
LAN extension.
Here are a few tutorials I gave on the topic: