https://github.com/sharathraparthy/nearest_sequence_memory
Reimplementation of the paper "Instance-Based State Identification for Reinforcement Learning "
https://github.com/sharathraparthy/nearest_sequence_memory
algorithm matlab reinforcement-learning reinforcement-learning-agent
Last synced: 8 months ago
JSON representation
Reimplementation of the paper "Instance-Based State Identification for Reinforcement Learning "
- Host: GitHub
- URL: https://github.com/sharathraparthy/nearest_sequence_memory
- Owner: SharathRaparthy
- Created: 2018-10-31T17:31:02.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2018-11-08T10:28:55.000Z (about 7 years ago)
- Last Synced: 2025-02-12T09:28:28.217Z (10 months ago)
- Topics: algorithm, matlab, reinforcement-learning, reinforcement-learning-agent
- Language: Matlab
- Homepage:
- Size: 43 KB
- Stars: 2
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Nearest Sequence Memory for Hidden State Idenification
This repository is the *reimplementation* of the paper "[Instance-Based State Identification for Reinforcement Learning](https://papers.nips.cc/paper/932-instance-based-state-identification-for-reinforcement-learning.pdf)" by [R.Andrew McCallum](https://people.cs.umass.edu/~mccallum/)
NSM is an instance-based algorithm for solving partially observable Markov decision problems (POMDPs). Here NSM algorithm is applies to a partially obsevable version of McCallum's grid-world presented in figure below.
### Prerequisites
```
Matlab 2015b (or later version)
Ubuntu 14.04 (or later version)/Windows
```
### Getting Started
After successful installation of matlab, clone this repository by using the following command
```
git clone https://github.com/SharathRaparthy/nearest_sequence_memory.git
```
Open your matlab and execute *rndTrial.m* script with the following MATLAB command:
```
plot(rndTrial(1000));
```
The result should like similar to figure below, but it should not be exactly the same.
Now run the NSMTrial function and plot the individual number of steps taken for 1000
episodes using the MATLAB command:
```
plot(NSMTrial(1000));
```
The result should like similar to Figure shown below but it should not be exactly the same.
