https://github.com/jshinm/deepnet-behavioral
Online behavioral experiment for deepnet paper
https://github.com/jshinm/deepnet-behavioral
behavioral-experiments extrapolation flask human-behavior machine-learning python
Last synced: about 1 month ago
JSON representation
Online behavioral experiment for deepnet paper
- Host: GitHub
- URL: https://github.com/jshinm/deepnet-behavioral
- Owner: jshinm
- License: mit
- Created: 2020-12-02T13:43:26.000Z (over 5 years ago)
- Default Branch: main
- Last Pushed: 2022-12-10T23:10:40.000Z (over 3 years ago)
- Last Synced: 2025-10-17T03:06:46.741Z (8 months ago)
- Topics: behavioral-experiments, extrapolation, flask, human-behavior, machine-learning, python
- Language: Jupyter Notebook
- Homepage:
- Size: 6.68 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Deepnet-behavioral
Online behavioral experiment for deepnet paper powered by Flask
https://ib-behavior.herokuapp.com/
The above website no longer available as heroku stopped free services as of Nov 28, 2022
## Installation
```
pip install -r requirements.txt
```
## Running Local Server
```
python main.py
```
## Logs
02/04/21
- [x] add codes for analytic pipeline
01/29/21
- [x] debug: sprial contracting issue
- [x] sample size reduced to 100
- [x] increase range from (-2,2) to (-3,+3)
- [x] number of trial increased from 50 to 100
- [x] add functionality to download sql db
- [x] minor adjustments to figure styling
01/08/21
- [x] debug: check for memory leakage (switched to "agg" backend for matplotlib)
- [x] connect to heroku postgres for database management (sqlite for local, postgres for production)
12/23/20
- [x] add "catch" trials (only sampled within the unit circle; interspersed throughout the test)
- [x] add practice trials at the end of the tutorials (currently 5 trials)
- [x] include consent language (placed in the very first page)
- [x] implement spiral posterior
- [x] spiral parameters modified (reduced noise 2.5 -> 1.0)
- [x] add feedback panel at the end of the experiment
- [ ] populate exactly the equal parts of each pattern
12/17/20
- [x] modify tutorial to emphasize likelihood judgement
- [x] further simplify tutorials and visually more attractive
- [x] remove slider bubble and change tick labels
12/09/20
- [x] Hard set the experiment trials to 50
- [x] Simplify tutorials
- [x] Create manual admin mode to activate/deactivate admin panels
- [x] Change slider style
- [x] Change overall interface styling
- [x] Update the manual
- [x] Have someone to run through the experiment
- [x] Admin mode deactivated for deployment