diff --git a/Dissertation.org b/Dissertation.org index 225fcdd..76861ed 100644 --- a/Dissertation.org +++ b/Dissertation.org @@ -8,6 +8,7 @@ #+PANDOC_OPTIONS: csl:assets/ieee.csl #+PANDOC_OPTIONS: pdf-engine:xelatex #+PANDOC_OPTIONS: top-level-division:chapter +#+PANDOC_OPTIONS: mathjax:t #+PANDOC_METADATA: link-citations:t * Resumen @@ -114,13 +115,27 @@ Una neurona artificial es un modelo de una neurona biológica, cada neurona reci #+NAME: fig:artificial-neuron [[./assets/figures/artificial-neuron.png]] -Una red neuronal artificial (ANN) es una red de capas de neuronas artificiales. Una ANN está formada por una capa de entrada, capas ocultas y una capa de salida. Las neuronas de una capa están conectadas, total o parcialmente, a las neuronas de la capa siguiente. También son posibles las conexiones de retroalimentación con las capas anteriores. La estructura típica de una ANN es la siguiente: +El proceso de activación se puede expresar como un modelo matemático: + +\begin{equation} +y= f \left(\sum\limits_{i=0}^{n} w_{i}x_{i} - T \right) +\end{equation} + +donde $y$ es la salida del nodo, $f$ es la función de activación, $w_i$ es el peso de la entrada $x_{i}$ , y $T$ es el valor del umbral. cite:Zou2009 + +Una red neuronal artificial (ANN) es una red de capas de neuronas artificiales. Una ANN está formada por una capa de entrada, capas ocultas y una capa de salida. Las neuronas de una capa están conectadas, total o parcialmente, a las neuronas de la capa siguiente. También son posibles las conexiones de retroalimentación con las capas anteriores. cite:book:80129 La estructura típica de una ANN es la siguiente: + +\clearpage #+CAPTION: Estructura de una red neuronal artificial cite:book:80129 #+ATTR_HTML: :height 30% :width 50% #+NAME: fig:neural-network [[./assets/figures/neural-network.png]] +Los principios básicos de las redes neuronales artificiales fueron formulados por primera vez en 1943, y el perceptrón, que históricamente es posiblemente la primera neurona artificial, se propuso en 1958. cite:book:2610592 Sin embargo, estos modelos no fueron populares hasta mediados de la década de 1980, cuando se reinventó el algoritmo de /back-propagation/. cite:book:771224 + +En la actualidad, los avances tanto en potencia de cálculo del /hardware/, especialmente en las tarjetas gráficas (GPU) cite:Cireşan2010, como la disponibilidad de grandes /datasets/ cite:book:771224 han dado lugar al /Deep Learning/. + ** Hacia el Deep Learning * Estado del arte ** Bioinformática diff --git a/Dissertation.pdf b/Dissertation.pdf index cc831dc..dfdc5dc 100644 Binary files a/Dissertation.pdf and b/Dissertation.pdf differ diff --git a/assets/bibliography.bib b/assets/bibliography.bib index f7941e7..04c14bd 100644 --- a/assets/bibliography.bib +++ b/assets/bibliography.bib @@ -561,7 +561,7 @@ year = 2010, series = {Prentice Hall Series in Artificial Intelligence}, edition = {3rd}, - pages = {38-45, 55-56} + pages = {38-45, 48-49, 55-56} } @article{McCarthy_Minsky_Rochester_Shannon_2006, @@ -599,3 +599,66 @@ edition = 2, pages = {39-40} } + +@Inbook{Zou2009, + author = "Zou, Jinming and Han, Yi and So, Sung-Sau", + editor = "Livingstone, David J.", + title = "Overview of Artificial Neural Networks", + bookTitle = "Artificial Neural Networks: Methods and Applications", + year = 2009, + publisher = "Humana Press", + address = "Totowa, NJ", + pages = "14--22", + abstract = "The artificial neural network (ANN), or simply neural + network, is a machine learning method evolved from the idea of + simulating the human brain. The data explosion in modern drug + discovery research requires sophisticated analysis methods to + uncover the hidden causal relationships between single or + multiple responses and a large set of properties. The ANN is + one of many versatile tools to meet the demand in drug + discovery modeling. Compared to a traditional regression + approach, the ANN is capable of modeling complex nonlinear + relationships. The ANN also has excellent fault tolerance and + is fast and highly scalable with parallel processing. This + chapter introduces the background of ANN development and + outlines the basic concepts crucially important for + understanding more sophisticated ANN. Several commonly used + learning methods and network setups are discussed briefly at + the end of the chapter.", + isbn = "978-1-60327-101-1", + doi = "10.1007/978-1-60327-101-1_2", + url = "https://doi.org/10.1007/978-1-60327-101-1_2" +} + +@book{book:2610592, + title = {Principles of artificial neural networks}, + author = {Graupe, Daniel}, + publisher = {World Scientific Publ}, + isbn = {9789814522731,9814522732}, + year = 2013, + edition = {3. ed}, + pages = {28-31} +} + +@Article{Cireşan2010, + author = {Cire{\c{s}}an, Dan Claudiu and Meier, Ueli and Gambardella, + Luca Maria and Schmidhuber, J{\"u}rgen}, + title = {Deep, Big, Simple Neural Nets for Handwritten Digit + Recognition}, + journal = {Neural Computation}, + year = 2010, + month = {Dec}, + day = 01, + volume = 22, + number = 12, + pages = {3207-3220}, + abstract = {Good old online backpropagation for plain multilayer + perceptrons yields a very low 0.35{\%} error rate on the MNIST + handwritten digits benchmark. All we need to achieve this best + result so far are many hidden layers, many neurons per layer, + numerous deformed training images to avoid overfitting, and + graphics cards to greatly speed up learning.}, + issn = {0899-7667}, + doi = {10.1162/NECO_a_00052}, + url = {https://doi.org/10.1162/NECO_a_00052} +}