diff --git a/README.md b/README.md index fca1612..6bc9860 100644 --- a/README.md +++ b/README.md @@ -1,16 +1,21 @@ -## Overview +# Redes neuronales y aprendiendo en lo profundo 🐱 -### neuralnetworksanddeeplearning.com integrated scripts for Python 3.5.2 and Theano with CUDA support +![https://img.shields.io/badge/license-MIT-green](https://img.shields.io/badge/license-MIT-green) -These scrips are updated ones from the **neuralnetworksanddeeplearning.com** gitHub repository in order to work with Python 3.5.2 -The testing file (**test.py**) contains all three networks (network.py, network2.py, network3.py) from the book and it is the starting point to run (i.e. *train and evaluate*) them. +## DescripciĂłn +En este repositorio vamos a aprender sobre redes neuronales, basandonos en el libro de Michael Nielsen [neuralnetworksanddeeplearning.com](http://neuralnetworksanddeeplearning.com/). Haremos videos explicativos sobre los diferentes temas basandonos en dichos capĂ­tulos del libro y mostraremos el funcionamiento de los cĂłdigos. -## Just type at shell: **python3.5 test.py** - -In test.py there are examples of networks configurations with proper comments. I did that to relate with particular chapters from the book. - -### License -Disributed under MIT License. [Link](LICENSE.md). +## CapĂ­tulos +1. **CapĂ­tulo 1**: "Usando redes neuronales para reconocer nĂșmeros escritos a mano". Veremos como una red neuronal puede aprender a reconocer dĂ­gitos escritos a mano, utilizando como ejemplo el conjunto de datos MNIST. Introduciremos conceptos bĂĄsicos sobre las redes neuronales. +2. **CapĂ­tulo 2**: "Como funciona el algoritmo de backpropagation". Analizaremos mĂĄs en profundidad, y utilizando matematicas la forma de aprender de las redes. +3. **CapĂ­tulo 3**: "Mejorando la forma en que las redes neuronales aprenden". Veremos algunas formas de mejorar la forma de aprendizaje de las redes neuronales. +4. **CapĂ­tulo 4**: "Una prueba grĂĄfica que las redes neuronales pueden computar cualquier funciĂłn". Estudiaremos y veremos de forma visual el teorema de Wierstrass. +5. **CapĂ­tulo 5**: "ÂżPorque las redes profundas son dificiles de entrenar?". Nos adentraremos en el mundo de las redes profundas, y por quĂ© son dĂ­ficiles de entrenar comparado con redes nueronales no profundas. +6. **CapĂ­tulo 6**: "Aprendizaje profundo". Mencionaremos brevemente el aprendizaje profundo de las redes neuronales neuronales. +## Reconocimientos +Este proyecto construye sobre un [fork de Michael Daniel Dobrzanski](https://github.com/MichalDanielDobrzanski/DeepLearningPython) del [proyecto original de Michael Nealsen](https://github.com/mnielsen/neural-networks-and-deep-learning). +## Licencia +Distribuido bajo una licencia MIT. [Link](LICENSE.md). diff --git a/e b/e new file mode 100644 index 0000000..52b7dfb --- /dev/null +++ b/e @@ -0,0 +1,132 @@ +commit 2da41b6dd19dafae92db1facbc4ed19738e80551 (HEAD -> master, origin/master, origin/HEAD) +Author: michaldobrzanski +Date: Mon Nov 27 14:07:08 2023 +0100 + + added MIT license + +commit 2eae26e0bdcef314dcb18f13946a94320fb28a12 +Merge: ea229ac 11cd4c1 +Author: MichaƂ Dobrzanski +Date: Wed Feb 24 08:57:56 2021 +0100 + + Merge pull request #32 from hamolicious/patch-1 + + Update network.py + +commit 11cd4c1643c39ada6a97b93dc2c522a3b59095ac +Author: Hamolicious <56944714+hamolicious@users.noreply.github.com> +Date: Fri Oct 23 13:12:15 2020 +0100 + + Update network.py + + - Removed semicolon line:73 + +commit ea229ac6234b7f3373f351f0b18616ca47edb8a1 +Merge: a8da42f 998e51b +Author: MichaƂ Dobrzanski +Date: Thu Apr 19 07:36:33 2018 +0200 + + Merge pull request #11 from dgcampbe/patch-1 + + Small update for Python 3 + +commit 998e51b741e95c2e6fd18cdba7df0c9b94e7957f +Author: Dane +Date: Fri Mar 2 17:04:26 2018 -0500 + + Add str() to variables being printed + + I realized that the variables weren't strings despite having "%s" (I thought the "s" meant string) in Python 2. That is what you get for only dealing with Python 3. :P + +commit ce2c044fb2b85371d65b4c71ab08b3bed9ba8067 +Author: Dane +Date: Fri Mar 2 01:42:31 2018 -0500 + + Small update for Python 3 + + Changed two print statements to print functions to be compatible with Python 3 instead of Python 2. + +commit a8da42fef60e9de977e3ad382c0804a1fe58cbc5 +Merge: f48defe 2a26d29 +Author: MichaƂ DobrzaƄski +Date: Mon Aug 28 11:01:48 2017 +0200 + + Merge pull request #2 from pa-m/master + + Avoid ImportError: cannot import name 'downsample'. New naming convention has been added in Theano 0.9 => use pool_2d instead + +commit 2a26d2937c2412c5ece5d26587a7351dcb226b67 +Author: pa-m +Date: Tue Apr 4 19:39:27 2017 +0200 + + Avoid ImportError: cannot import name 'downsample' + Since a change in theano 0.9 downsample.max_pool_2d has to be replaced with pool.pool_2d. + +commit f48defe06bfa1127fbda4f054b27d1e23c17c725 +Merge: e99416e 3ef0f0b +Author: MichaƂ DobrzaƄski +Date: Thu Feb 9 10:31:15 2017 +0100 + + Merge pull request #1 from anandman/master + + updated mnist_average_darkness.py for Python 3.5.2 - further improvements + +commit 3ef0f0b575e0a6a96508fbb9ae6c51be2def8672 +Author: anand m +Date: Sat Feb 4 13:58:16 2017 -0800 + + updated for Python 3.5.2 + +commit e99416e6ad59e602d0dc3709a445c3a4efad259f +Author: MichaƂ DobrzaƄski +Date: Sun Oct 23 12:00:59 2016 +0200 + + Update README.md + +commit a35f06c85e2903c81e3f9394ad7e5f24443e40ae +Author: MichaƂ DobrzaƄski +Date: Sun Oct 23 11:57:30 2016 +0200 + + Working on network3.py - convolutional NN examples added + +commit 664aee31f815956aaef96a533521d1eddd5fb283 +Author: MichaƂ DobrzaƄski +Date: Sun Sep 25 16:56:45 2016 +0200 + + Working Theano with CUDA7.5 Support at Ubuntu 16.04 + +commit f403c099829ee70700b60569a546ab7e9d6793b9 +Author: MichaƂ DobrzaƄski +Date: Wed Sep 14 19:46:01 2016 +0200 + + Working on network3.py and Theano with CUDA + +commit 092007f9158a4a3ff56a08d03250a19ae8953753 +Author: MichaƂ DobrzaƄski +Date: Tue Sep 13 11:51:49 2016 +0200 + + Chapter 4 - vanishing gradient problem in a deep networks + +commit 7283525363f379b99779c96b6cc82b25626c2249 +Author: MichaƂ DobrzaƄski +Date: Fri Sep 9 15:28:59 2016 +0200 + + Early stopping implemented at network2.py + +commit 021657efa2f874453ebf94ca1598b38add9b7dc5 +Author: MichaƂ DobrzaƄski +Date: Sat Jul 30 18:08:54 2016 +0200 + + Updated readme + 2nd example from 3rd chapter + +commit 988965a43e7b002c5bc03ecb675e9c0f15ba9072 +Author: MichaƂ DobrzaƄski +Date: Fri Jul 29 20:23:55 2016 +0200 + + Create README.md + +commit 0edab8e2501621752d6dab41bf193a380eb2819b +Author: MichaƂ DobrzaƄski +Date: Fri Jul 29 20:06:48 2016 +0200 + + Initial files