Text Normalization(文本正则化)

Overview

Text Normalization(文本正则化)

任务描述:通过机器学习算法将英文文本的“手写”形式转换成“口语“形式,例如“6ft”转换成“six feet”等

实验结果

  1. XGBoost + bag-of-words: 0.99159
  2. XGBoost+Weights+rules:0.99002
  3. 进阶solve函数(使用6个output文件):0.98939
  4. 基本solve函数:0.98277
  5. RandomTree + Rules:0.95304
  6. XGboost:0.92605

参考github网址:

数据来源网址:

数据分析EDA网址(帮助快速理解数据特征):

提升点

1. 数据的不平衡性

对于平衡的数据,我们一般使用准确率作为一般的评估标准(accuracy),当类别不平衡时,准确率就具有迷惑性,而且意义不大。因此有以下几种主流评测标准

  • Receiver operating curve,计算ROC曲线面积(二分类,从PLAIN和非PLAIN)
  • Precision-recall curve,计算此曲线下的面积
  • Precision

- 简单通用的算法

阈值调整(threshold moving):将原本默认为0.5的阈值调整到 较少类别/(较少类别+较多类别)即可。使用现有的集成学习分类器,如随机森林或者xgboost,并调整分类阈值

- 对XGBoost模型数据的不平衡处理方法

通过正负样本的权重解决样本不均衡(一般分类中小样本量类别权重高,大样本类别权重低,再进行计算和建模

- 简单有效的方案

  1. 不对数据进行过采样和欠采样,但使用现有的集成学习模型,如随机森林,XGBoost(lGBM)
  2. 输出模型的预测概率,调整阈值得到最终结果
  3. 选择合适的评估标准,如precision,Recall
  4. 文本正则化中的任务是对测试集中的16个目标进行预测,训练集中的最大类别是PLAIN,为7353693,最小的类别为ADDRESS,为522。因此暂定PLAIN的权重为0.01,其余为1.(除去PLAIN,其余15个再做一次分类)

2. 超参数优化(时间复杂度,空间复杂度)

如何选择合适的超参数?不同模型会有不同的最优超参数组合,找到这组最优超参数大家是根据经验或者随机的方法,来尝试。但是其是有可能用数学或者机器学习的模型来解决模型本身超参数的选择问题

背景

  • 机器学习模型超参数调优一般被认为是一个黑盒优化问题,在调优过程中我们只能看到模型的输入与输出,不能获取模型训练过程中的梯度信息,也不能假设模型超参数和最终指标符合凸优化条件
  • 模型训练代价大,时间,金钱成本

自动调参方法

Grid search(网格搜索),Random search(随机搜索),Genetic algorithm(遗传算法),Paticle Swarm Optimization(粒子群优化),Bayesian Optimization(贝叶斯优化),TPE,SMAC等

  • Genetic algorithm和PSO是经典黑盒优化算法,归类为群体优化算法,不是特别适合模型超参数调优场景,因为其需要有足够多的初始样本点,并且优化效率不高**
  • Grid search很容易理解与实现,但是遍历所有的超参数组合来找到其中最优化的方案,对于连续值还需要等间距采样。实际上这30种组合不一定取得全局最优解,而且计算量很大很容易组合爆炸,并不是一种高效的参数调优方法。
  • Random search普遍被认为比Grid search效果好,虽然组合的超参数具有随机性,但是其出现效果可能特别差也可能特别好,在尝试次数和Grid search相同的情况下一般最值会更大,当然variance也更大但这不影响最终结果。

但是在计算机资源有限的情况下,Grid search与Random search不一定比建模工程师的经验要好

  • Bayesian Optimization
    适用场景:
    (1)需要优化的function计算起来非常费时费力,比如上面提到的神经网络的超参问题,每一次训练神经网络都是燃烧好多GPU的
    (2)你要优化的function没有导数信息

3. 可解释性工具(https://www.kaggle.com/learn/machine-learning-explainability)

Xgboost相对于线性模型在进行预测时往往有更好的精度,但是同时也失去了线性模型的可解释性。所以Xgboost通常被认为是黑箱模型。 经典方法是使用全局特征重要性评估

2017年,Lundberg和Lee的[论文]( [A Unified Approach to Interpreting Model Predictions.pdf](../文献阅读/A Unified Approach to Interpreting Model Predictions.pdf) )提出了SHAP值这一广泛适用的方法用来解释各种模型(分类以及回归),其中最大的受益者莫过于之前难以被理解的黑箱模型,如boosting和神经网络模型。

  1. 二分类,看下准确率,高的话
  2. 集成XGBoost,LGB,随机森林
  3. 可解释性,SHAP(SHAP值只能对特征进行分析)
  4. 去掉PLAIN看下效果,ROC

后续改进

  1. 将PLAIN的权值设置为0,训练结果:分数为0.98991 将PLAIN去除不进行预测,实验结果无法得到官方分数,并且实验是通过上下文单词(context)来作为单位进行训练,若去除PLAIN无法训练。因此只能通过将权值设置为0,查看各个种类的预测准确率是否高,但是可以查看对训练集的效果
Owner
Jason_Zhang
Jason_Zhang
Materials (slides, code, assignments) for the NYU class I teach on NLP and ML Systems (Master of Engineering).

FREE_7773 Repo containing material for the NYU class (Master of Engineering) I teach on NLP, ML Sys etc. For context on what the class is trying to ac

Jacopo Tagliabue 90 Dec 19, 2022
Applying "Load What You Need: Smaller Versions of Multilingual BERT" to LaBSE

smaller-LaBSE LaBSE(Language-agnostic BERT Sentence Embedding) is a very good method to get sentence embeddings across languages. But it is hard to fi

Jeong Ukjae 13 Sep 02, 2022
Pipelines de datos, 2021.

Este repo ilustra un proceso sencillo de automatización de transformación y modelado de datos, a través de un pipeline utilizando Luigi. Stack princip

Rodolfo Ferro 8 May 19, 2022
German Text-To-Speech Engine using Tacotron and Griffin-Lim

jotts JoTTS is a German text-to-speech engine using tacotron and griffin-lim. The synthesizer model has been trained on my voice using Tacotron1. Due

padmalcom 6 Aug 28, 2022
Asr abc - Automatic speech recognition(ASR),中文语音识别

语音识别的简单示例,主要在课堂演示使用 创建python虚拟环境 在linux 和macos 上验证通过 # 如果已经有pyhon3.6 环境,跳过该步骤,使用

LIyong.Guo 8 Nov 11, 2022
Code for papers "Generation-Augmented Retrieval for Open-Domain Question Answering" and "Reader-Guided Passage Reranking for Open-Domain Question Answering", ACL 2021

This repo provides the code of the following papers: (GAR) "Generation-Augmented Retrieval for Open-domain Question Answering", ACL 2021 (RIDER) "Read

morning 49 Dec 26, 2022
Natural Language Processing Specialization

Natural Language Processing Specialization In this folder, Natural Language Processing Specialization projects and notes can be found. WHAT I LEARNED

Kaan BOKE 3 Oct 06, 2022
InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective

InfoBERT: Improving Robustness of Language Models from An Information Theoretic Perspective This is the official code base for our ICLR 2021 paper

AI Secure 71 Nov 25, 2022
A model library for exploring state-of-the-art deep learning topologies and techniques for optimizing Natural Language Processing neural networks

A Deep Learning NLP/NLU library by Intel® AI Lab Overview | Models | Installation | Examples | Documentation | Tutorials | Contributing NLP Architect

Intel Labs 2.9k Jan 02, 2023
Code for CVPR 2021 paper: Revamping Cross-Modal Recipe Retrieval with Hierarchical Transformers and Self-supervised Learning

Revamping Cross-Modal Recipe Retrieval with Hierarchical Transformers and Self-supervised Learning This is the PyTorch companion code for the paper: A

Amazon 69 Jan 03, 2023
T‘rex Park is a Youzan sponsored project. Offering Chinese NLP and image models pretrained from E-commerce datasets

T‘rex Park is a Youzan sponsored project. Offering Chinese NLP and image models pretrained from E-commerce datasets (product titles, images, comments, etc.).

55 Nov 22, 2022
A PyTorch Implementation of End-to-End Models for Speech-to-Text

speech Speech is an open-source package to build end-to-end models for automatic speech recognition. Sequence-to-sequence models with attention, Conne

Awni Hannun 647 Dec 25, 2022
A very simple framework for state-of-the-art Natural Language Processing (NLP)

A very simple framework for state-of-the-art NLP. Developed by Humboldt University of Berlin and friends. IMPORTANT: (30.08.2020) We moved our models

flair 12.3k Dec 31, 2022
Pytorch version of BERT-whitening

BERT-whitening This is the Pytorch implementation of "Whitening Sentence Representations for Better Semantics and Faster Retrieval". BERT-whitening is

Weijie Liu 255 Dec 27, 2022
Predict the spans of toxic posts that were responsible for the toxic label of the posts

toxic-spans-detection An attempt at the SemEval 2021 Task 5: Toxic Spans Detection. The Toxic Spans Detection task of SemEval2021 required participant

Ilias Antonopoulos 3 Jul 24, 2022
Code for text augmentation method leveraging large-scale language models

HyperMix Code for our paper GPT3Mix and conducting classification experiments using GPT-3 prompt-based data augmentation. Getting Started Installing P

NAVER AI 47 Dec 20, 2022
CCKS-Title-based-large-scale-commodity-entity-retrieval-top1

- 基于标题的大规模商品实体检索top1 一、任务介绍 CCKS 2020:基于标题的大规模商品实体检索,任务为对于给定的一个商品标题,参赛系统需要匹配到该标题在给定商品库中的对应商品实体。 输入:输入文件包括若干行商品标题。 输出:输出文本每一行包括此标题对应的商品实体,即给定知识库中商品 ID,

43 Nov 11, 2022
Mastering Transformers, published by Packt

Mastering Transformers This is the code repository for Mastering Transformers, published by Packt. Build state-of-the-art models from scratch with adv

Packt 195 Jan 01, 2023
One Stop Anomaly Shop: Anomaly detection using two-phase approach: (a) pre-labeling using statistics, Natural Language Processing and static rules; (b) anomaly scoring using supervised and unsupervised machine learning.

One Stop Anomaly Shop (OSAS) Quick start guide Step 1: Get/build the docker image Option 1: Use precompiled image (might not reflect latest changes):

Adobe, Inc. 148 Dec 26, 2022
构建一个多源(公众号、RSS)、干净、个性化的阅读环境

2C 构建一个多源(公众号、RSS)、干净、个性化的阅读环境 作为一名微信公众号的重度用户,公众号一直被我设为汲取知识的地方。随着使用程度的增加,相信大家或多或少会有一个比较头疼的问题——广告问题。 假设你关注的公众号有十来个,若一个公众号两周接一次广告,理论上你会面临二十多次广告,实际上会更多,运

howie.hu 678 Dec 28, 2022