murraryzhao / adaptive-decision-boundary Goto Github PK

A method to automatically learn the adaptive decision boundary (ADB) for open world classification.

The proposed method together with baselines are also integrated into the open intent detection module in our another scalable framework TEXTOIR, enjoy it!

This repository provides the official PyTorch implementation of the research paper Deep Open Intent Classification with Adaptive Decision Boundary (Accepted by AAAI2021).

Related works can refer to the reading list.

We use anaconda to create python environment:

conda create --name python=3.6

Install all required libraries:

pip install -r requirements.txt

Download the pre-trained bert model (bert-base-uncased) from the following link:
Baidu Cloud Drive with code: v8tk

Set the path of the uncased-bert model (parameter "bert_model" in init_parameter.py).

Run the experiments by:

sh scripts/run.sh

You can change the parameters in the script. The selected parameters are as follows:

dataset: clinc | banking | oos (default)
known_class_ratio: 0.25 | 0.5 | 0.75 (default)
labeled_ratio: 0.2 | 0.4 | 0.6 | 0.8 | 1.0 (default) 

The model architecture of ADB:

The detailed results can be seen in results.md.

*KIR means "Known Intent Ratio".

“Open” and “Known” denote the macro f1-score over open class and known classes respectively.

If you are insterested in this work, and want to use the codes or results in this repository, please star this repository and cite by:

@article{Zhang_Xu_Lin_2021, 
      title={Deep Open Intent Classification with Adaptive Decision Boundary}, 
      volume={35}, 
      number={16}, 
      journal={Proceedings of the AAAI Conference on Artificial Intelligence}, 
      author={Zhang, Hanlei and Xu, Hua and Lin, Ting-En}, 
      year={2021}, 
      month={May}, 
      pages={14374-14382} 
}

This paper is founded by seed fund of Tsinghua University (Department of Computer Science and Technology)- Siemens Ltd., China Joint Research Center for Industrial Intelligence and Internet of Things.