diff --git a/contributing.md b/contributing.md index f8624ac..f7b67d0 100644 --- a/contributing.md +++ b/contributing.md @@ -9,9 +9,9 @@ Our goal is to have a list of things that anybody would have to learn if they we Your contributions to this repo are always welcome! Bear in mind, that this repo is *highly opinionated*, *unbiased* and *curated*. -Your opinion on value of any resource may not match the opinion of curator. +Your opinion on value of any resource may not match the opinion of the curators. -**No PR will be discarded without explanations!** +**No PR will be discarded without explanation!** ## How are these roadmaps made? @@ -23,12 +23,12 @@ Your opinion on value of any resource may not match the opinion of curator. * Export the xml file `File > Export as > XML > uncheck "compressed" > Export` and put it in the `images` directory * Export the svg file `File > Export as > SVG > uncheck "Include a copy of my diagram" > Export` and put it in the `images` directory * Make sure that the width and height of the SVG match the actual image content to avoid unnecessary whitespace around the image -* Commits your changes and open a PR +* Commit your changes and open a PR ## Guidelines * **Adding everything available out there is not the goal!** -The roadmaps represents the skillset most valuable today i.e. if you were to enter any of the listed fields today, what would you learn! There might be things that are of-course being used today but prioritize the things that are most in demand today. Use your critical thinking to filter out non-essential stuff. Give honest arguments for why the resource should be included. +The roadmaps represent the skillset most valuable in present time meaning that if you were to enter any of the listed fields today, that's what you would learn first. Although there is an endless amount of techniques, tools and components that were important in the past or appear promising to become relevant in the future, prioritize the skills that are generally most relevant for today's demand. Use your critical thinking to filter out non-essential stuff. Give honest arguments for why the resource should be included. * **One item per Pull Request** There may be a discussion related to an item you want to add. Adding just a single item per pull request makes it much easier for everyone involved. * Write meaningful commit messages diff --git a/images/deep_learning.svg b/images/deep_learning.svg index efe9744..095b40a 100644 --- a/images/deep_learning.svg +++ b/images/deep_learning.svg @@ -1,3 +1,3 @@ -

Deep Learning

Deep Learning
Deep Learning Papers Reading Roadmap
Deep Learning Papers Reading...
Papers
Papers
Papers with code
Papers with code
Papers with code - state of the art
Papers with code - state of...
Understanding
Neural Networks
Understanding...
Neural Networks
Neural Networks
Perceptrons
Perceptrons
Autoencoders
Autoencoders
Convolutional Neural Networks
(CNN)
Convolutional Neural Network...
Generative Adversarial Networks (GAN)
Generative Adversarial Netwo...
Architectures
Architectures
Awesome Deep Learning
Awesome Deep Learning
Tools
Tools
PyTorch
PyTorch

keep exploring and stay up-to-date

keep exploring and s...
Recurrent Neural Networks
(RNN)
Recurrent Neural Networks...
LSM
LSM
LSTM
LSTM
GRU
GRU
Tensorflow
Tensorflow
Viewer does not support full SVG 1.1
\ No newline at end of file +

Deep Learning

Deep Learning
Deep Learning Papers Reading Roadmap
Deep Learning Papers Reading...
Papers
Papers
Papers with code
Papers with code
Papers with code - state of the art
Papers with code - state of...
Understanding
Neural Networks
Understanding...
Neural Networks
Neural Networks
Feedforward neural network
Feedforward neural network
Autoencoder
Autoencoder
Convolutional Neural Network
(CNN)
Convolutional Neural Network...
Generative Adversarial Network (GAN)
Generative Adversarial Netwo...
Architecture
Architecture
Awesome Deep Learning
Awesome Deep Learning
Tools
Tools
PyTorch
PyTorch

keep exploring and stay up-to-date

keep exploring and s...
Recurrent Neural Network
(RNN)
Recurrent Neural Network...
LSTM
LSTM
GRU
GRU
Tensorflow
Tensorflow
Loss Functions
Loss Functions
Activation Functions
Activation Functions
Weight Initialization
Weight Initialization
Vanishing / Exploding
Gradient Problem
Vanishing / Exploding...
Pooling
Pooling
Transformer
Transformer
Encoder
Encoder
Decoder
Decoder
Attention
Attention
Siamese Network
Siamese Network
Residual Connections
Residual Connections
Optimizers
Optimizers
Training
Training
Learning Rate Schedule
Learning Rate Schedule
Batch Normalization
Batch Normalization
Batch Size Effects
Batch Size Effects
Regularization
Regularization
Multitask Learning
Multitask Learning
Transfer Learning
Transfer Learning
Curriculum Learning
Curriculum Learning
SGD
SGD
Momentum
Momentum
Adam
Adam
AdaGrad
AdaGrad
AdaDelta
AdaDelta
Nadam
Nadam
RMSProp
RMSProp
Early Stopping
Early Stopping
Dropout
Dropout
Parameter Penalties
Parameter Penalties
Data Augmentation
Data Augmentation
Adversarial Training
Adversarial Training
Tensorboard
Tensorboard
MLFlow
MLFlow
Distillation
Distillation
Model optimization
(advanced)
Model optimiza...
Neural Architecture
Search (NAS)
Neural Architecture...
Quantization
Quantization
Viewer does not support full SVG 1.1
\ No newline at end of file diff --git a/images/deep_learning.xml b/images/deep_learning.xml index 81e8987..640a140 100644 --- a/images/deep_learning.xml +++ b/images/deep_learning.xml @@ -1,6 +1,7 @@ - + + - + @@ -8,29 +9,29 @@ - + - - + + - - + + - + - + - + @@ -38,107 +39,403 @@ - + - + - + - + - + - + - + - - - - + + + + + + - + - + - - - - - + + + + + + + - + - + - - + + + + - + - + - + - + - + - - - - - - - - - + + + - - - + + + + + - + - + - + - + - + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +