AEPsych

AEPsych

  • Docs
  • Tutorials
  • API Reference
  • GitHub

›General

About

  • Introduction
  • Papers related to AEPsych

General

  • Getting Started with the AEPsych service
  • Writing Config Files
  • AEPsych clients

Background materials

  • A brief introduction to Psychophysics
  • A brief introduction to Gaussian Process active learning

For developers

  • API Overview
  • Database Overview

Advanced topics

  • Advanced Strategy Configuration
  • The Ax Backend
Edit

Getting Started with the AEPsych service

Installation

AEPsych only supports python 3.8+. We recommend installing AEPsych under a virtual environment like Anaconda. Once you've created a virtual environment for AEPsych and activated it, you can install AEPsych from pip:

pip install aepsych

This installs the core library, as well as our executable aepsych_server. For installing clients, see client documentation.

The AEPsych experiment flow

Here is the basic overview of how AEPsych works: assets/flowchart.png

First, we initialize a model from a small number of datapoints collected from quasi-random parameter settings. Then:

  1. Build a model of the data so far. AEPsych uses Gaussian Process classification models by default, but other models can be used as well if they are API-compatible with AEPsych and differentiable using PyTorch.
  2. Using the model to select the next point to observe. AEPsych does this by defining an acquisition function that describes the goodness of observing a particular point given the knowledge encoded in the model (for example, the uncertainty in the model's predictions at that point), and then optimizing that function w.r.t. the stimulus configuration.
  3. Query the participant for a response. Here AEPsych can interface with PsychoPy, PsychToolbox, or other stimulus presentation code.
  4. Go back to 1.
← Papers related to AEPsychWriting Config Files →
  • Installation
  • The AEPsych experiment flow
AEPsych
Docs
IntroductionGetting StartedTutorialsAPI Reference
Legal
PrivacyTerms
Social
aepsych
Facebook Open Source
Copyright © 2023 Meta, Inc. Built with Docusaurus.

We use cookies to enhance your experience, and to analyse the use of our website. By clicking or navigating, you agree to allow our usage of cookies.