MARE - Multi-Attribute Relation Extraction

Related tags

Deep Learningmare
Overview

MARE - Multi-Attribute Relation Extraction

Repository for the paper submission: #TODO: insert link, when available

Environment

Tested with Ubuntu 18.04, Anaconda 2020.11 and NVIDIA driver version 450.102.04 If you have a lower driver version and you don't need to train own models, we recommend to install the environment without the cuda requirement.

If you have no coda-compatible GPU, delete the cudatoolkit dependency from the environment.yml file. If you do not have a cuda GPU, remove the line

- cudatoolkit=10.2.89

from environment.yml.

To install the conda environment execute

conda env create -f environment.yml

Install mare (the local directory) via pip

pip install -e .

This may take several minutes.

Reproduction of results

To reproduce the values from the Paper, download the corresponding models from https://fh-aachen.sciebo.de/s/D5FLVN7qk2UTCmX and put the .tar.gz files in the models folder or execute the following shell commands.

wget -c https://fh-aachen.sciebo.de/s/D5FLVN7qk2UTCmX/download -O models.zip

unzip models.zip

rm models.zip

The following instructions can be used to reproduce the results in the paper. All evaluations create a subfolder in evaluations.

Sequence Tagging

The values for AR, Cl, MRE, CRE und BRE correspond to the values of MARE Seq. Tag. in Table 2.

The F1 scores for AR_no_trigger und MRE_no_trigger correspond to the values of Seq. Tag. with Trigger un Table 3

sh scripts/evaluate_model.sh models/sequence.tar.gz evaluations/seq_tag seq_lab_elmo_pred mare

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.42957746478873204
recall_micro: 0.4765625
f1_micro: 0.45185185185185106

EVALUATION RESULTS FOR Cl

precision_micro: 0.725352112676056
recall_micro: 0.8046875
f1_micro: 0.7629629629629631

EVALUATION RESULTS FOR CRE

precision_micro: 0.28169014084507005
recall_micro: 0.3125
f1_micro: 0.296296296296296

EVALUATION RESULTS FOR AR

precision_micro: 0.660412757973733
recall_micro: 0.6591760299625461
f1_micro: 0.659793814432989

EVALUATION RESULTS FOR BRE

precision_micro: 0.439252336448598
recall_micro: 0.49473684210526303
f1_micro: 0.46534653465346504

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.464788732394366
recall_micro: 0.515625
f1_micro: 0.48888888888888804

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.6410891089108911
recall_micro: 0.6301703163017031
f1_micro: 0.635582822085889

Span Labeling

The values for AR, Cl, MRE, CRE und BRE correspond to the values of MARE Span Lab. in Table 2.

The F1 scores for AR_no_trigger und MRE_no_trigger correspond to the values of Span Lab. with Trigger un Table 3

sh scripts/evaluate_model.sh models/span_based.tar.gz evaluations/span_lab mare.span_based_precidtor.SpanBasedPredictor mare

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.47244094488188904
recall_micro: 0.46875000000000006
f1_micro: 0.47058823529411703

EVALUATION RESULTS FOR Cl

precision_micro: 0.8031496062992121
recall_micro: 0.796875
f1_micro: 0.8

EVALUATION RESULTS FOR CRE

precision_micro: 0.291338582677165
recall_micro: 0.2890625
f1_micro: 0.290196078431372

EVALUATION RESULTS FOR AR

precision_micro: 0.751619870410367
recall_micro: 0.651685393258427
f1_micro: 0.698094282848545

EVALUATION RESULTS FOR BRE

precision_micro: 0.49473684210526303
recall_micro: 0.49473684210526303
f1_micro: 0.49473684210526303

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.519685039370078
recall_micro: 0.515625
f1_micro: 0.517647058823529

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.7298850574712641
recall_micro: 0.618004866180048
f1_micro: 0.6693017127799731

Dygie ++

The values for AR, Cl, MRE, CRE und BRE correspond to the values of Dygie++ in Table 2.

The F1 scores for AR_no_trigger und MRE_no_trigger correspond to the values of Dygie++ with Trigger in Table 3

sh scripts/evaluate_model.sh models/dygiepp.tar.gz evaluations/dygiepp mare.evaluation.mock_model.DygieppMockModel mare

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.47154471544715404
recall_micro: 0.453125
f1_micro: 0.46215139442231

EVALUATION RESULTS FOR Cl

precision_micro: 0.7723577235772351
recall_micro: 0.7421875
f1_micro: 0.7569721115537841

EVALUATION RESULTS FOR CRE

precision_micro: 0.260162601626016
recall_micro: 0.25
f1_micro: 0.254980079681274

EVALUATION RESULTS FOR AR

precision_micro: 0.630434782608695
recall_micro: 0.651685393258427
f1_micro: 0.6408839779005521

EVALUATION RESULTS FOR BRE

precision_micro: 0.550561797752809
recall_micro: 0.51578947368421
f1_micro: 0.5326086956521741

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.536585365853658
recall_micro: 0.515625
f1_micro: 0.525896414342629

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.596810933940774
recall_micro: 0.6374695863746951
f1_micro: 0.616470588235294

SpERT (SpART = SpERT with AllenNLP)

The value for BRE corresponds to the values of SpERT in Table 2.

sh scripts/evaluate_model.sh models/spart.tar.gz evaluations/spert spart spart

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.43269230769230704
recall_micro: 0.3515625
f1_micro: 0.387931034482758

EVALUATION RESULTS FOR Cl

precision_micro: 0.596153846153846
recall_micro: 0.484375
f1_micro: 0.5344827586206891

EVALUATION RESULTS FOR CRE

precision_micro: 0.08653846153846101
recall_micro: 0.0703125
f1_micro: 0.077586206896551

EVALUATION RESULTS FOR AR

precision_micro: 0.519230769230769
recall_micro: 0.202247191011235
f1_micro: 0.2911051212938

EVALUATION RESULTS FOR BRE

precision_micro: 0.573333333333333
recall_micro: 0.45263157894736805
f1_micro: 0.505882352941176

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.48076923076923006
recall_micro: 0.390625
f1_micro: 0.43103448275862005

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.528
recall_micro: 0.16058394160583903
f1_micro: 0.246268656716417

Sequence Tagging Baseline

The values for AR, Cl, MRE, CRE und BRE correspond to the values of MARE Baseline in Table 2.

sh scripts/evaluate_model.sh models/sequence_tagging_baseline.tar.gz evaluations/seq_tag_baseline seq_lab_elmo_pred mare

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.396825396825396
recall_micro: 0.390625
f1_micro: 0.39370078740157405

EVALUATION RESULTS FOR Cl

precision_micro: 0.682539682539682
recall_micro: 0.671875
f1_micro: 0.677165354330708

EVALUATION RESULTS FOR CRE

precision_micro: 0.26190476190476103
recall_micro: 0.2578125
f1_micro: 0.259842519685039

EVALUATION RESULTS FOR AR

precision_micro: 0.6591422121896161
recall_micro: 0.5468164794007491
f1_micro: 0.597748208802456

EVALUATION RESULTS FOR BRE

precision_micro: 0.40206185567010305
recall_micro: 0.410526315789473
f1_micro: 0.40625000000000006

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.42857142857142805
recall_micro: 0.421875
f1_micro: 0.42519685039370003

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.6296296296296291
recall_micro: 0.49635036496350304
f1_micro: 0.5551020408163261

Sequence Tagging No Trigger

The F1 scores for AR_no_trigger und MRE_no_trigger correspond to the values of Seq. Tag. without Trigger in Table 3

Change the include_trigger Parameter in mare/seq_lab_elmo_pred.py to False.

sh scripts/evaluate_model.sh models/sequence_no_trigger.tar.gz evaluations/seq_tag_no_trig seq_lab_elmo_pred_no_trig mare

The result shoud be


EVALUATION RESULTS FOR MRE

precision_micro: 0.056
recall_micro: 0.0546875
f1_micro: 0.055335968379446

EVALUATION RESULTS FOR Cl

precision_micro: 0.728
recall_micro: 0.7109375
f1_micro: 0.7193675889328061

EVALUATION RESULTS FOR CRE

precision_micro: 0.048
recall_micro: 0.046875
f1_micro: 0.047430830039525

EVALUATION RESULTS FOR AR

precision_micro: 0.662337662337662
recall_micro: 0.47752808988764006
f1_micro: 0.554951033732317

EVALUATION RESULTS FOR BRE

precision_micro: 0.07865168539325801
recall_micro: 0.073684210526315
f1_micro: 0.07608695652173901

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.512
recall_micro: 0.5
f1_micro: 0.50592885375494

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.662337662337662
recall_micro: 0.620437956204379
f1_micro: 0.6407035175879391

Span Labeling No Trigger

The F1 scores for AR_no_trigger und MRE_no_trigger correspond to the values of Span Lab. without Trigger in Table 3

sh scripts/evaluate_model.sh models/span_based_no_trigger_local.tar.gz evaluations/span_lab_no_trig mare.span_based_precidtor.SpanBasedPredictor mare

The result shoud be

EVALUATION RESULTS FOR MRE

precision_micro: 0.07563025210084001
recall_micro: 0.0703125
f1_micro: 0.072874493927125

EVALUATION RESULTS FOR Cl

precision_micro: 0.789915966386554
recall_micro: 0.734375
f1_micro: 0.761133603238866

EVALUATION RESULTS FOR CRE

precision_micro: 0.067226890756302
recall_micro: 0.0625
f1_micro: 0.064777327935222

EVALUATION RESULTS FOR AR

precision_micro: 0.72
recall_micro: 0.47191011235955005
f1_micro: 0.570135746606334

EVALUATION RESULTS FOR BRE

precision_micro: 0.103448275862068
recall_micro: 0.09473684210526301
f1_micro: 0.09890109890109801

EVALUATION RESULTS FOR MRE_no_trigger

precision_micro: 0.5630252100840331
recall_micro: 0.5234375
f1_micro: 0.542510121457489

EVALUATION RESULTS FOR AR_no_trigger

precision_micro: 0.72
recall_micro: 0.613138686131386
f1_micro: 0.6622864651773981
Owner
GitHub Repository
Melanoma Skin Cancer Detection using Convolutional Neural Networks and Transfer LearningðŸ•ĩðŸŧ‍♂ïļ

This is a Kaggle competition in which we have to identify if the given lesion image is malignant or not for Melanoma which is a type of skin cancer.

Vipul Shinde 1 Jan 27, 2022
Awesome Long-Tailed Learning

Awesome Long-Tailed Learning This repo pays specially attention to the long-tailed distribution, where labels follow a long-tailed or power-law distri

Stomach_ache 284 Jan 06, 2023
Code for CVPR2021 paper "Learning Salient Boundary Feature for Anchor-free Temporal Action Localization"

AFSD: Learning Salient Boundary Feature for Anchor-free Temporal Action Localization This is an official implementation in PyTorch of AFSD. Our paper

Tencent YouTu Research 146 Dec 24, 2022
Official Pytorch Implementation of Relational Self-Attention: What's Missing in Attention for Video Understanding

Relational Self-Attention: What's Missing in Attention for Video Understanding This repository is the official implementation of "Relational Self-Atte

mandos 43 Dec 07, 2022
Official implementation of Self-supervised Graph Attention Networks (SuperGAT), ICLR 2021.

SuperGAT Official implementation of Self-supervised Graph Attention Networks (SuperGAT). This model is presented at How to Find Your Friendly Neighbor

Dongkwan Kim 127 Dec 28, 2022
Ultra-lightweight human body posture key point CNN model. ModelSize:2.3MB HUAWEI P40 NCNN benchmark: 6ms/img,

Ultralight-SimplePose Support NCNN mobile terminal deployment Based on MXNET(=1.5.1) GLUON(=0.7.0) framework Top-down strategy: The input image is t

223 Dec 27, 2022
HyperSeg: Patch-wise Hypernetwork for Real-time Semantic Segmentation Official PyTorch Implementation

: We present a novel, real-time, semantic segmentation network in which the encoder both encodes and generates the parameters (weights) of the decoder. Furthermore, to allow maximal adaptivity, the w

Yuval Nirkin 182 Dec 14, 2022
Rational Activation Functions - Replacing PadÃĐ Activation Units

Rational Activations - Learnable Rational Activation Functions First introduce as PAU in PadÃĐ Activation Units: End-to-end Learning of Activation Func

<a href=[email protected]"> 38 Nov 22, 2022
IDM: An Intermediate Domain Module for Domain Adaptive Person Re-ID,

Intermediate Domain Module (IDM) This repository is the official implementation for IDM: An Intermediate Domain Module for Domain Adaptive Person Re-I

Yongxing Dai 87 Nov 22, 2022
Official implementation of "A Unified Objective for Novel Class Discovery", ICCV2021 (Oral)

A Unified Objective for Novel Class Discovery This is the official repository for the paper: A Unified Objective for Novel Class Discovery Enrico Fini

Enrico Fini 118 Dec 26, 2022
Event sourced bank - A wide-and-shallow example using the Python event sourcing library

Event Sourced Bank A "wide but shallow" example of using the Python event sourci

3 Mar 09, 2022
Research - dataset and code for 2016 paper Learning a Driving Simulator

the people's comma the paper Learning a Driving Simulator the comma.ai driving dataset 7 and a quarter hours of largely highway driving. Enough to tra

comma.ai 4.1k Jan 02, 2023
Official pytorch implementation of DeformSyncNet: Deformation Transfer via Synchronized Shape Deformation Spaces

DeformSyncNet: Deformation Transfer via Synchronized Shape Deformation Spaces Minhyuk Sung*, Zhenyu Jiang*, Panos Achlioptas, Niloy J. Mitra, Leonidas

Zhenyu Jiang 21 Aug 30, 2022
PyTorch Connectomics: segmentation toolbox for EM connectomics

Introduction The field of connectomics aims to reconstruct the wiring diagram of the brain by mapping the neural connections at the level of individua

Zudi Lin 132 Dec 26, 2022
Let's create a tool to convert Thailand budget from PDF to CSV.

thailand-budget-pdf2csv Let's create a tool to convert Thailand Government Budgeting from PDF to CSV! āļĢāļ§āļĄāļžāļĨāļąāļ‡ Dev āđāļ›āļĨāļ‡āļ‡āļš āļˆāļēāļ PDF āļŠāļđāđˆ Machine-readable

Kao.Geek 88 Dec 19, 2022
KAPAO is an efficient multi-person human pose estimation model that detects keypoints and poses as objects and fuses the detections to predict human poses.

KAPAO (Keypoints and Poses as Objects) KAPAO is an efficient single-stage multi-person human pose estimation model that models keypoints and poses as

Will McNally 664 Dec 30, 2022
Neural-net-from-scratch - A simple Neural Network from scratch in Python using the Pymathrix library

A Simple Neural Network from scratch A Simple Neural Network from scratch in Pyt

Youssef Chafiqui 2 Jan 07, 2022
Towards Calibrated Model for Long-Tailed Visual Recognition from Prior Perspective

Towards Calibrated Model for Long-Tailed Visual Recognition from Prior Perspective Zhengzhuo Xu, Zenghao Chai, Chun Yuan This is the PyTorch implement

Sincere 16 Dec 15, 2022
Rust bindings for the C++ api of PyTorch.

tch-rs Rust bindings for the C++ api of PyTorch. The goal of the tch crate is to provide some thin wrappers around the C++ PyTorch api (a.k.a. libtorc

Laurent Mazare 2.3k Dec 30, 2022
TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-Captured Scenarios

TPH-YOLOv5 This repo is the implementation of "TPH-YOLOv5: Improved YOLOv5 Based on Transformer Prediction Head for Object Detection on Drone-Captured

cv516Buaa 439 Dec 22, 2022