Abstract
  1. Principal Mel-Spectrum Components (Feature)
  2. Temporal Pooling Functions (Model)
  3. Single Hidden Layer Neural Network, thus Multi-layer Perceptron (Classifier)

Audio Preprocessing
    Feature: PMSC (Principal Mel-Spectrum Components)
  1. Original Data:  30s, 22.05KHz, mono, wav
  2. Process Steps:
    1. DFT (spectral domain)
      we compute DFTs over windows of 1024 samples on audio at 22.05 KHz (i.e. roughly 46ms) with a frame step of 512 samples.
    2. Mel-Compression
      we run the spectral amplitudes through a set of 256 mel-scaled triangular filters to abtain a set of spectral energy bands.
    3. Principal Component analysis whitening (PCA whitening)
      we compute the principal components of a random sub-sample of training set. In order to obtain features with unitary variance, we multiply(乘以) each component by the inverse square of its eigenvalue(特征值平方的倒数). ---- PCA whitening.
Model
    PFC (Pooled Features Classifier)
  1. Pooling Operation
    the model applies a given set of pooling functions (how many?) to the PMSC features, and sends the pooled features to a classifier(MLP, with hidden layer of 2000 units, sigmoid activation, L2 weight decay and cross-entropy cost).
  2. Classify
    each pooling window is considered as a training example for the classifier, and average the predictions of the classifier over all the windows of a given clip to obtain the final classification (what is the rule?).
Tasks
  1. Classification (train/test task)
    the MLP outputs an affinity prediction for each class (pooling functions tread each pooling window as a training example).
  2. Tagging
    1. Affinity
      the affinity scores for a song is thus directly the output of the MLP.
    2. Binary Classification
      choose the threshold that optimizes the F1-score on the validation set.
Tools
  1. Theano: Theano is a numerical computation library for Python. In Theano, computations are expressed using a NumPy-like syntax and compiled to run efficiently on either CPU or GPU architectures.
    


PH_Pooled Featrues Classification MIREX 2011 Submission的更多相关文章

  1. 2013:Audio Tag Classification - MIREX Wiki

    Contents [hide] 1 Description 1.1 Task specific mailing list 2 Data 2.1 MajorMiner Tag Dataset 2.2 M ...

  2. 2011:Audio Classification (Train/Test) Tasks - MIREX Wiki

    Contents [hide] 1 Audio Classification (Test/Train) tasks 1.1 Description 1.1.1 Task specific mailin ...

  3. Metric Learning度量学习:**矩阵学习和图学习

    DML学习原文链接:http://blog.csdn.net/lzt1983/article/details/7884553 一篇metric learning(DML)的综述文章,对DML的意义.方 ...

  4. Compile For Cydia Submission Author: BigBoss Updated September 23, 2011

    Compile For Cydia Submission Author: BigBoss Updated September 23, 2011: In order to submit your app ...

  5. [MIREX] MIREX评测介绍

    MIREX作为国际最权威音频检索评测大赛,竟然在百度上找不到任何介绍,只有几个与什么搜狗.腾讯获得什么成绩相关的检索内容,相比而言,TRECVID的内容收到重视多了...由于研究生阶段主要研究音频领域 ...

  6. Twin Prime Conjecture(浙大计算机研究生保研复试上机考试-2011年)

    Twin Prime Conjecture                                            Time Limit: 2000/1000 MS (Java/Othe ...

  7. HDU 4041 Eliminate Witches! (模拟题 ACM ICPC 2011亚洲北京赛区网络赛)

    HDU 4041 Eliminate Witches! (模拟题 ACM ICPC 2011 亚洲北京赛区网络赛题目) Eliminate Witches! Time Limit: 2000/1000 ...

  8. HDU 4046 Panda (ACM ICPC 2011北京赛区网络赛)

    HDU 4046 Panda (ACM ICPC 2011北京赛区网络赛) Panda Time Limit: 10000/4000 MS (Java/Others)    Memory Limit: ...

  9. A Novel Multi-label Classification Based on PCA and ML-KNN

     ICIC Express Letters                  ICIC International ⓒ2010 ISSN 1881-803X Volume4, Number5, O ...

随机推荐

  1. Webpack 入门(一):安装 / 打包 / 命令行

    一:安装webpack和基本环境搭建 新建一个工作的文件夹(我取的名字叫Webpack) 打开命令行,cd进入该文件夹 //初始化一下npm > E:\work\Webpack>npm i ...

  2. 富文本编辑器复制Wod字体问题

    目前常用的富文本编辑器:百度版UEditor,wangEditor,ckeditor,kindeditor,TinyMCE.当Word复制文本粘贴到编辑器时,几乎都无法保证字体大小完全一致的问题. 想 ...

  3. Log.d 日志调试查看(所有平台)

    https://www.cnblogs.com/onechen/p/6436748.html http://docwiki.embarcadero.com/Libraries/Berlin/en/FM ...

  4. 奇异值分解 SVD 的数学解释

    奇异值分解(Singular Value Decomposition,SVD)是一种矩阵分解(Matrix Decomposition)的方法.除此之外,矩阵分解还有很多方法,例如特征分解(Eigen ...

  5. SCOI2013 密码

    题目描述: Fish是一条生活在海里的鱼.有一天他很无聊,就到处去寻宝.他找到了位于海底深处的宫殿,但是一扇带有密码锁的大门却阻止了他的前进. 通过翻阅古籍,Fish 得知了这个密码的相关信息: 该密 ...

  6. [Python3网络爬虫开发实战] 1.9.1-Docker的安装

    Docker是一种容器技术,可以将应用和环境等进行打包,形成一个独立的.类似于iOS的App形式的“应用”.这个应用可以直接被分发到任意一个支持Docker的环境中,通过简单的命令即可启动运行.Doc ...

  7. python各种推导式分析

    推导式comprehensions(又称解析式),是Python的一种独有特性.推导式是可以从一个数据序列构建另一个新的数据序列的结构体. 共有三种推导,在Python2和3中都有支持: 列表(lis ...

  8. LeetCode(47)Permutations II

    题目 Given a collection of numbers that might contain duplicates, return all possible unique permutati ...

  9. BUAA_OO_博客作业四

    BUAA_OO_博客作业四 1 第四单元两次作业的架构设计 1.1 第13次作业 类图 ​ 作业要求:通过实现UmlInteraction这个官方提供的接口,来实现自己的UmlInteraction解 ...

  10. 前端跳转处理--房天下的访问页面部分ip自动跳转到登录页面的解决办法(xjl456852原创)

    朋友说自己在访问房天下的页面时,他们页面进行了跳转,跳转到登录页面,说是前端跳转.让我也看看,我看我的机器没有进行跳转. 后来就发现有的机器在访问页面会自动跳转到登录页面.有的不会进行跳转. 比如访问 ...