LSTM for HAR
MXNet-scala module implementation of LSTM for Human Activity Recognition.
Based on: https://github.com/guillaume-chevalier/LSTM-Human-Activity-Recognition
and got nearly the same results.
Tested on Ubuntu 14.04
sbt 0.13
Mxnet
1, compile Mxnet with CUDA, then compile the scala-pkg;
2, cd into Mxnet-Scala/HumanActivityRecognition, then mkdir lib;
3, copy your compiled mxnet-full_2.11-linux-x86_64-gpu-0.9.5-SNAPSHOT.jar into lib folder;
4, run sbt, compile the project
cd scripts;
bash run.sh
then have fun!
Iter 1500, Batch Loss = 1.780415, Accuracy = 0.26266667 TEST SET DISPLAY STEP: Batch Loss = 1.798351, Accuracy = 0.16016288 Iter 15000, Batch Loss = 0.933169, Accuracy = 0.33733332 TEST SET DISPLAY STEP: Batch Loss = 1.017054, Accuracy = 0.35934848 Iter 30000, Batch Loss = 1.203503, Accuracy = 0.364 TEST SET DISPLAY STEP: Batch Loss = 1.210487, Accuracy = 0.41940957 Iter 45000, Batch Loss = 0.740827, Accuracy = 0.43066666 TEST SET DISPLAY STEP: Batch Loss = 0.772413, Accuracy = 0.43094674 Iter 60000, Batch Loss = 0.852083, Accuracy = 0.412 TEST SET DISPLAY STEP: Batch Loss = 0.893236, Accuracy = 0.3790295 Iter 75000, Batch Loss = 0.823337, Accuracy = 0.426 TEST SET DISPLAY STEP: Batch Loss = 0.811571, Accuracy = 0.4183916 Iter 90000, Batch Loss = 0.753601, Accuracy = 0.518 TEST SET DISPLAY STEP: Batch Loss = 0.760808, Accuracy = 0.49677637 Iter 105000, Batch Loss = 0.655924, Accuracy = 0.5326667 TEST SET DISPLAY STEP: Batch Loss = 0.705918, Accuracy = 0.5551408 Iter 120000, Batch Loss = 0.671545, Accuracy = 0.62133336 TEST SET DISPLAY STEP: Batch Loss = 0.700519, Accuracy = 0.5846624 Iter 135000, Batch Loss = 0.520920, Accuracy = 0.64133334 TEST SET DISPLAY STEP: Batch Loss = 0.509846, Accuracy = 0.60366476 Iter 150000, Batch Loss = 0.529982, Accuracy = 0.628 TEST SET DISPLAY STEP: Batch Loss = 0.541752, Accuracy = 0.586359 Iter 165000, Batch Loss = 0.337939, Accuracy = 0.66933334 TEST SET DISPLAY STEP: Batch Loss = 0.333877, Accuracy = 0.6671191 Iter 180000, Batch Loss = 0.384434, Accuracy = 0.6626667 TEST SET DISPLAY STEP: Batch Loss = 0.414936, Accuracy = 0.68069226 Iter 195000, Batch Loss = 0.369616, Accuracy = 0.71533334 TEST SET DISPLAY STEP: Batch Loss = 0.388370, Accuracy = 0.672209 Iter 210000, Batch Loss = 0.355249, Accuracy = 0.64533335 TEST SET DISPLAY STEP: Batch Loss = 0.391691, Accuracy = 0.6789956 Iter 225000, Batch Loss = 0.431139, Accuracy = 0.57533336 TEST SET DISPLAY STEP: Batch Loss = 0.441554, Accuracy = 0.6973193 Iter 240000, Batch Loss = 0.285340, Accuracy = 0.84066665 TEST SET DISPLAY STEP: Batch Loss = 0.328950, Accuracy = 0.75500506 Iter 255000, Batch Loss = 0.289774, Accuracy = 0.8513333 TEST SET DISPLAY STEP: Batch Loss = 0.307293, Accuracy = 0.7804547 Iter 270000, Batch Loss = 0.276257, Accuracy = 0.8846667 TEST SET DISPLAY STEP: Batch Loss = 0.299869, Accuracy = 0.8154055 Iter 285000, Batch Loss = 0.221921, Accuracy = 0.92466664 TEST SET DISPLAY STEP: Batch Loss = 0.242042, Accuracy = 0.79911774 Iter 300000, Batch Loss = 0.195684, Accuracy = 0.908 TEST SET DISPLAY STEP: Batch Loss = 0.199364, Accuracy = 0.8191381 Iter 315000, Batch Loss = 0.165992, Accuracy = 0.92266667 TEST SET DISPLAY STEP: Batch Loss = 0.172706, Accuracy = 0.8269427 Iter 330000, Batch Loss = 0.130108, Accuracy = 0.9253333 TEST SET DISPLAY STEP: Batch Loss = 0.135752, Accuracy = 0.8259247 Iter 345000, Batch Loss = 0.123394, Accuracy = 0.92733335 TEST SET DISPLAY STEP: Batch Loss = 0.162172, Accuracy = 0.85103494 Iter 360000, Batch Loss = 0.091235, Accuracy = 0.9173333 TEST SET DISPLAY STEP: Batch Loss = 0.124131, Accuracy = 0.7896166 Iter 375000, Batch Loss = 0.110845, Accuracy = 0.9166667 TEST SET DISPLAY STEP: Batch Loss = 0.158513, Accuracy = 0.8408551 Iter 390000, Batch Loss = 0.096496, Accuracy = 0.922 TEST SET DISPLAY STEP: Batch Loss = 0.134032, Accuracy = 0.85714287 Iter 405000, Batch Loss = 0.147098, Accuracy = 0.91933334 TEST SET DISPLAY STEP: Batch Loss = 0.192178, Accuracy = 0.82422805 Iter 420000, Batch Loss = 0.128252, Accuracy = 0.918 TEST SET DISPLAY STEP: Batch Loss = 0.143951, Accuracy = 0.8517136 Iter 435000, Batch Loss = 0.113734, Accuracy = 0.8986667 TEST SET DISPLAY STEP: Batch Loss = 0.135311, Accuracy = 0.8439091 Iter 450000, Batch Loss = 0.110106, Accuracy = 0.94 TEST SET DISPLAY STEP: Batch Loss = 0.124025, Accuracy = 0.8449271 Iter 465000, Batch Loss = 0.127751, Accuracy = 0.928 TEST SET DISPLAY STEP: Batch Loss = 0.131118, Accuracy = 0.8561249 Iter 480000, Batch Loss = 0.112045, Accuracy = 0.9433333 TEST SET DISPLAY STEP: Batch Loss = 0.117775, Accuracy = 0.86868 Iter 495000, Batch Loss = 0.100415, Accuracy = 0.956 TEST SET DISPLAY STEP: Batch Loss = 0.111091, Accuracy = 0.8805565 Iter 510000, Batch Loss = 0.092717, Accuracy = 0.95266664 TEST SET DISPLAY STEP: Batch Loss = 0.106292, Accuracy = 0.8754666 Iter 525000, Batch Loss = 0.078547, Accuracy = 0.956 TEST SET DISPLAY STEP: Batch Loss = 0.106027, Accuracy = 0.8788599 Iter 540000, Batch Loss = 0.097594, Accuracy = 0.88733333 TEST SET DISPLAY STEP: Batch Loss = 0.101987, Accuracy = 0.87987787 Iter 555000, Batch Loss = 0.094006, Accuracy = 0.8386667 TEST SET DISPLAY STEP: Batch Loss = 0.094256, Accuracy = 0.8866644 Iter 570000, Batch Loss = 0.074015, Accuracy = 0.87866664 TEST SET DISPLAY STEP: Batch Loss = 0.091538, Accuracy = 0.8846284 Iter 585000, Batch Loss = 0.092351, Accuracy = 0.8653333 TEST SET DISPLAY STEP: Batch Loss = 0.094474, Accuracy = 0.8819138 Iter 600000, Batch Loss = 0.087074, Accuracy = 0.862 TEST SET DISPLAY STEP: Batch Loss = 0.089067, Accuracy = 0.88394976 Iter 615000, Batch Loss = 0.072428, Accuracy = 0.93266666 TEST SET DISPLAY STEP: Batch Loss = 0.081364, Accuracy = 0.88394976 Iter 630000, Batch Loss = 0.063190, Accuracy = 0.9533333 TEST SET DISPLAY STEP: Batch Loss = 0.075417, Accuracy = 0.8866644 Iter 645000, Batch Loss = 0.072255, Accuracy = 0.982 TEST SET DISPLAY STEP: Batch Loss = 0.091810, Accuracy = 0.8819138 Iter 660000, Batch Loss = 0.069323, Accuracy = 0.954 TEST SET DISPLAY STEP: Batch Loss = 0.085984, Accuracy = 0.88632506 Iter 675000, Batch Loss = 0.063361, Accuracy = 0.98066664 TEST SET DISPLAY STEP: Batch Loss = 0.093897, Accuracy = 0.87410927 Iter 690000, Batch Loss = 0.047272, Accuracy = 0.996 TEST SET DISPLAY STEP: Batch Loss = 0.074714, Accuracy = 0.89786226 Iter 705000, Batch Loss = 0.037897, Accuracy = 0.97333336 TEST SET DISPLAY STEP: Batch Loss = 0.071891, Accuracy = 0.90091616 Iter 720000, Batch Loss = 0.048627, Accuracy = 0.948 TEST SET DISPLAY STEP: Batch Loss = 0.086670, Accuracy = 0.8856464 Iter 735000, Batch Loss = 0.036653, Accuracy = 0.948 TEST SET DISPLAY STEP: Batch Loss = 0.061301, Accuracy = 0.88734305 FINAL RESULT: Batch Loss= 0.061300557, Accuracy= 0.88734305 Done
链接:https://github.com/Ldpe2G/DeepLearningForFun/tree/master/Mxnet-Scala/HumanActivityRecognition
上一篇:Self-Norm Nets
下一篇:pix2pix
还没有评论,说两句吧!
热门资源
seetafaceJNI
项目介绍 基于中科院seetaface2进行封装的JAVA...
spark-corenlp
This package wraps Stanford CoreNLP annotators ...
Keras-ResNeXt
Keras ResNeXt Implementation of ResNeXt models...
capsnet-with-caps...
CapsNet with capsule-wise convolution Project ...
inferno-boilerplate
This is a very basic boilerplate example for pe...
智能在线
400-630-6780
聆听.建议反馈
E-mail: support@tusaishared.com