發(fā)簡信
IP屬地：上海
-
論文筆記-Convolutional Neural Networks for Speech Recognition問題：ASR里用CNN做聲學(xué)模型伏钠，輸入特征FBANK钩杰，采用三通道形式作為輸入，請問如何處理句子不同幀數(shù)問題苛萎？ https://www.micro...
-
論文筆記-SPEAKER RECOGNITION FROM RAW WAVEFORM WITH SINCNETDNNs have also been proposed for direct discrimnative speaker classifica...
-
senet <-incepetion,resnet<-vggDeeper architectures https://arxiv.org/pdf/1709.01507.pdf 關(guān)鍵詞 learning 桨昙，...
-
senet論文筆記參考文章： 解讀Squeeze-and-Excitation Networks（SENet）SENet學(xué)習(xí)筆記Squeeze-and-Excit...
-
tensorflow template機(jī)器學(xué)習(xí)的三個核心的定義：定義模型、定義目標(biāo)函數(shù)和定義優(yōu)化方法 構(gòu)建計算圖 分發(fā)計算任務(wù) 執(zhí)行計算任務(wù)另外還要準(zhǔn)備數(shù)據(jù) 對大多數(shù)的任務(wù)來說腌歉，第2...
-
inception系列論文摘錄(v1,v2,v3)論文： https://arxiv.org/pdf/1409.4842v1.pdfhttps://arxiv.org/pdf/1502.0316...
-
論文摘錄:resnet&batch normalization
Deep Residual Learning for Image Recognition 開門見山蛙酪，拋問題 問題1、An obstacle to...
-
speaker論文摘錄
Deep Speaker: an End-to-End Neural Speaker Embedding System - 5 May 201...
-
tensorflow 離線安裝問題集
html5lib cannot be found in bleach installation Tensorflow不同版本要求與CUDA及CU...