bowdar / Deeplearning
Neural network base on c++14, support any number of layers 基于C++14元编程的深度学习神经网络模板类,支持任意层数
Stars: ✭ 80
Programming Languages
metaprogramming
66 projects
Labels
Projects that are alternatives of or similar to Deeplearning
Ncnn Benchmark
The benchmark of ncnn that is a high-performance neural network inference framework optimized for the mobile platform
Stars: ✭ 70 (-12.5%)
Mutual labels: deeplearning
Tools To Design Or Visualize Architecture Of Neural Network
Tools to Design or Visualize Architecture of Neural Network
Stars: ✭ 1,143 (+1328.75%)
Mutual labels: deeplearning
Deep learning for biologists with keras
tutorials made for biologists to learn deep learning
Stars: ✭ 74 (-7.5%)
Mutual labels: deeplearning
Bidaf Keras
Bidirectional Attention Flow for Machine Comprehension implemented in Keras 2
Stars: ✭ 60 (-25%)
Mutual labels: deeplearning
Skip Thought Tf
An implementation of skip-thought vectors in Tensorflow
Stars: ✭ 77 (-3.75%)
Mutual labels: deeplearning
Blinkdl
A minimalist deep learning library in Javascript using WebGL + asm.js. Run convolutional neural network in your browser.
Stars: ✭ 69 (-13.75%)
Mutual labels: deeplearning
Deeplearning.ai Notes
基于Andrew Ng DeepLearning.ai课程的学习笔记
Stars: ✭ 75 (-6.25%)
Mutual labels: deeplearning
Pwc Net
PWC-Net: CNNs for Optical Flow Using Pyramid, Warping, and Cost Volume, CVPR 2018 (Oral)
Stars: ✭ 1,142 (+1327.5%)
Mutual labels: deeplearning
Ssd Models
把极速检测器的门槛给我打下来make lightweight caffe-ssd great again
Stars: ✭ 62 (-22.5%)
Mutual labels: deeplearning
Udacity Natural Language Processing Nanodegree
Tutorials and my solutions to the Udacity NLP Nanodegree
Stars: ✭ 73 (-8.75%)
Mutual labels: deeplearning
Cnn Paper2
🎨 🎨 深度学习 卷积神经网络教程 :图像识别,目标检测,语义分割,实例分割,人脸识别,神经风格转换,GAN等🎨🎨 https://dataxujing.github.io/CNN-paper2/
Stars: ✭ 77 (-3.75%)
Mutual labels: deeplearning
Mit Deep Learning
Tutorials, assignments, and competitions for MIT Deep Learning related courses.
Stars: ✭ 8,912 (+11040%)
Mutual labels: deeplearning
Meta-programming DeepLearning
Meta-programming neural network 是一个基于C++14实现的元编程神经网络库 Compile-time matrix constructions, headonly, no dependency, limitless layers, limitless nodes
Feature
- 支持任意深度和超大结点数
- 矩阵运算(CNN采用张量运算)
- 循环类网络输入输出支持多对单、单对多、多对多
- 源码Head-only并且无依赖
- 使用方法极其简单,适合程序局部应用ANN以及用来学习研究
Sample
1) BPNN
#include "BPNN.hpp"
int main()
{
/// 1. Create a 4 layers NN each layer nodes are 20, 30, 20 and 2
/// The first 20 is input layer and the last 2 is output
typedef mtl::BPNN<20, 30, 20, 2> MyNN;
MyNN bpnn;
/// 2. Initialize, setup parameters and activate functions
bpnn.init()
.set_aberration(0.0001)
.set_learnrate(0.8)
.set_sigfunc(mtl::logsig)
.set_dsigfunc(mtl::dlogsig);
/// 3. Create input output matrixs, and then enter matrix datas your self
MyNN::InMatrix inMx;
MyNN::OutMatrix outMx;
MyNN::OutMatrix expectMx;
/// enter matrix datas ...
/// 4. Training, call train in your own way
bpnn.train(inMx, outMx, 100);
/// 5. Simulate
bpnn.simulate(inMx, outMx, expectMx);
}
2) RNN
#include "RNN.hpp"
int main()
{
/// 1. Create a 4 layers NN each layer nodes are 20, 30, 20 and 2
/// The first 20 is input layer and the last 2 is output
typedef mtl::RNN<20, 30, 20, 2> MyRnn;
MyRnn rnn;
/// 2. Initialize, setup parameters and activate functions
rnn.init()
.set_aberration(0.0001)
.set_learnrate(0.8)
.set_sigfunc(mtl::logsig)
.set_dsigfunc(mtl::dlogsig);
/// 3. Create input output matrixs, and then enter matrix datas your self
/// RNN suport multi-in-out like M:1, 1:M and M:M also 1:1 which is meaningless
MyRnn::InMatrix<10> inMx; /// 10 input a group, you can change it each training
MyRnn::OutMatrix<2> outMx; /// 2 ouput a group
MyRnn::OutMatrix<2> expectMx;
/// enter matrix datas ...
/// 4. Training, call train in your own way
rnn.train(inMx, outMx, 100);
/// 5. Simulate
rnn.simulate(inMx, outMx,expectMx);
}
3) LSTM
#include "LSTM.hpp"
int main()
{
/// 1. Create a 4 layers NN each layer nodes are 20, 30, 20 and 2
/// The first 20 is input layer and the last 2 is output
typedef mtl::LSTM<20, 30, 20, 2> MyLSTM;
MyLSTM lstm;
/// 2. Initialize, setup parameters, LSTM wouldn't setup activate functions
lstm.init()
.set_aberration(0.0001)
.set_learnrate(0.8);
/// 3. Create input output matrixs, and then enter matrix datas your self
/// RNN suport multi-in-out like M:1, 1:M and M:M also 1:1 which is meaningless
MyLSTM::InMatrix<10> inMx;
MyLSTM::OutMatrix<2> outMx;
MyLSTM::OutMatrix<2> expectMx;
/// enter matrix datas ...
/// 4. Training, call train in your own way
lstm.train(inMx, outMx, 100);
/// 5. Simulate
lstm.simulate(inMx, outMx,expectMx);
}
Note that the project description data, including the texts, logos, images, and/or trademarks,
for each open source project belongs to its rightful owner.
If you wish to add or remove any projects, please contact us at [email protected].