Model Visualization

Visualization

OJS

Published

January 9, 2024

Code

#lapply(c('tidyverse','data.table','igraph','ggraph','kableExtra'),library,character.only=TRUE))
pacman::p_load(tidyverse,data.table,igraph,ggraph,kableExtra,DiagrammeR,png, plantuml)

plantuml visualizations

Code

uml <- '@startuml

skinparam ranksep 20
skinparam dpi 125
skinparam packageTitleAlignment left

rectangle "Input Layer" {
  (Input1)
  (Input2)
}

rectangle "Output Layer" {
  (Output1)
  (Output2)
  (Output3)
}

rectangle "<b>ALM (Associative Learning Model)</b>" as ALM_model

(Input1) ..> (Output1) : w_{11}
(Input1) ..> (Output2) : w_{12}
(Input1) ..> (Output3) : w_{13}

(Input2) ..> (Output1) : w_{21}
(Input2) ..> (Output2) : w_{22}
(Input2) ..> (Output3) : w_{23}

ALM_model ==> (Input Layer)
(Input Layer) --> (Output Layer)

@enduml
'
uml <- plantuml(uml)
plot(x=uml)

[1] "/var/folders/r1/c92xp9t13094kv05vft2mbzw0000gn/T//RtmpDQ2JVS/plantuml.e77d28bb8fbd.svg"

Code

alm_model_with_equations <- '
@startuml

skinparam ranksep 20
skinparam dpi 125
skinparam packageTitleAlignment left

rectangle "Input Layer" {
  (Input1)
  (Input2)
}

rectangle "Output Layer" {
  (Output1)
  (Output2)
  (Output3)
}


(Input1) ..> (Output1) : w_{11}
(Input1) ..> (Output2) : w_{12}
(Input1) ..> (Output3) : w_{13}

(Input2) ..> (Output1) : w_{21}
(Input2) ..> (Output2) : w_{22}
(Input2) ..> (Output3) : w_{23}


rectangle "<b>ALM (Associative Learning Model)</b>" as ALM_model

ALM_model ==> (Input Layer)
(Input Layer) --> (Output Layer)

note right of (Input Layer)
  <math>a_i(X)=e^{-\\gamma \\cdot(X-X_i)^2}</math>
  <math>Normalized: \\frac{a_i(X)}{\\sum a_i(X)}</math>
end note

note right of (Output Layer)
  blank
end note

@enduml
'

# Pass the PlantUML code to plantuml function
alm_model_with_equations <- plantuml(alm_model_with_equations)
plot(x=alm_model_with_equations)

[1] "/var/folders/r1/c92xp9t13094kv05vft2mbzw0000gn/T//RtmpDQ2JVS/plantuml.e77d547cc13a.svg"

ggplot model visualization

Code

nInput=6
nOutput=10

inNodes <- seq(1,nInput,1) %>% as.integer()
outNodes <- seq(300,1000,length.out=nOutput)%>% as.integer()
weight.mat <<- matrix(0.001,nrow=nOutput,ncol=nInput) # weights initialized to 0 (as in Delosh 1997)

stim <- 1.5
c=.1
inAct <- round(exp(-c*((inNodes-stim)^2)),2)
inActLab <- paste0("x",inNodes,"=",inAct)
outAct <- weight.mat %*% inAct
output.probability <<- outAct/sum(outAct)
outLab=paste0("y",outNodes,"=",round(output.probability,2))
mean.response <<- round(sum(outNodes * output.probability),0)


resp <- mean.response
inFlow <- tibble(expand.grid(from=stim,to=inActLab)) %>% mutate_all(as.character)
outFlow <- tibble(expand.grid(from=outLab,to=mean.response)) %>% mutate_all(as.character)

gd <- tibble(expand.grid(from=inActLab,to=outLab)) %>% mutate_all(as.character) %>%
  rbind(inFlow,.) %>% rbind(.,outFlow)

xInc <- .3
yInc=.5

g = graph_from_data_frame(gd,directed=TRUE)
coords2=layout_as_tree(g)
colnames(coords2)=c("y","x")

odf <- as_tibble(coords2) %>% 
  mutate(label=vertex_attr(g,"name"),
         type=c("stim",rep("Input",nInput),rep("Output",nOutput),"Resp"),
         x=x*-1) %>%
  mutate(y=ifelse(type=="Resp",0,y),xmin=x-xInc,xmax=x+xInc,ymin=y-yInc,ymax=y+yInc)

plot_edges = gd %>% mutate(id=row_number()) %>%
  pivot_longer(cols=c("from","to"),names_to="s_e",values_to=("label")) %>%
                 mutate(label=as.character(label)) %>% 
  group_by(id) %>%
  mutate(weight=sqrt(rnorm(1,mean=0,sd=10)^2)/10) %>%
  left_join(odf,by="label") %>%
  mutate(xmin=xmin+.02,xmax=xmax-.02)

ggplot() + geom_rect(data = odf,
            mapping = aes(xmin = xmin, ymin = ymin, 
                          xmax = xmax, ymax = ymax, 
                          fill = type, colour = type),alpha = 0.01) +
  geom_text(data=odf,aes(x=x,y=y,label=label,size=3)) +
  geom_path(data=plot_edges,mapping=aes(x=x,y=y,group=id,alpha=weight)) +
  # geom_rect(aes(xmin=-1.05,xmax=-.95,ymin=-10,ymax=5),color="red",alpha=.1)+
  # geom_rect(aes(xmin=-0.05,xmax=.05,ymin=-10,ymax=5),color="blue",alpha=.1) +
  theme_void()

ggraph method

Code

inNodes <- seq(1,6,1) %>% as.integer()
outNodes <- seq(300,1000,50)%>% as.integer()

da <- data.frame(expand.grid(inNodes,outNodes))  %>% magrittr::set_colnames(c("input","output"))
da <- da %>% mutate_all(as.character)
m = graph_from_data_frame(da, directed = TRUE)

coords = layout_with_sugiyama(m)
colnames(coords$layout) = c("y", "x")
coords$layout=coords$layout[,c("x","y")]
plot(m,layout=coords)

Code

ggraph(m,layout=coords$layout)+
geom_edge_link0(width=0.2,colour="grey")+
  geom_node_point(col="white",size=6)+scale_x_reverse()+
  geom_node_text(aes(label=name)) +
  # draw rectangle that covers input layer at x=1, min y is min of coords$y and max y is max of coords$y
  annotate("rect",xmin=0,xmax=.1,ymin=min(coords$layout[,2]),ymax=max(coords$layout[,2]),fill="grey",alpha=0.7)

Code

 geom_rect(xmin=0,xmax=1.1,ymin=min(coords$layout[,2]),ymax=max(coords$layout[,2]),fill="grey",alpha=0.7)

geom_rect: linejoin = mitre, na.rm = FALSE
stat_identity: na.rm = FALSE
position_identity

Code

library(tidyverse)
library(ggforce)

theme_set(theme_grey() +
            theme_void() +
            theme(plot.margin = margin(0, 5.5, 0, 5.5)))

# Parameters
input_x <- c(1, 2)
output_x <- c(1.5, 2.5, 3.5)
input_y <- 3
output_y <- 1
activation_levels <- c(0.2, 0.7, 0.5)

# Input Layer
input_layer <- tibble(x = input_x, y = rep(input_y, length(input_x)))

# Output Layer
output_layer <- tibble(x = output_x, y = rep(output_y, length(output_x)), activation = activation_levels)

# Gaussian Activation
gaussian_activation <- tibble(
  x = seq(from = 0, to = 3, by = 0.01),
  y = exp(-2 * (x - 1)^2) + 2.5,
  x2 = x,
  y2 = exp(-2 * (x - 2)^2) + 2.5
)

# Plot
p <- ggplot() +
  # Input nodes
  geom_point(data = input_layer, aes(x = x, y = y), size = 4, color = 'green') +
  annotate("text", x = input_x, y = rep(input_y, length(input_x)) + 0.3, label = c("Input #1", "Input #2")) +
  
  # Gaussian Activations
  geom_line(data = gaussian_activation, aes(x = x, y = y), color = 'blue') +
  geom_line(data = gaussian_activation, aes(x = x2, y = y2), color = 'blue') +
  
  # Connections
  geom_segment(data = expand.grid(input_x, output_x), aes(x = Var1, xend = Var2, y = input_y, yend = output_y), arrow = arrow(type = 'closed', length = unit(0.2, 'inches'))) +
  
  # Output nodes
  geom_point(data = output_layer, aes(x = x, y = y), size = 4, color = 'red') +
  geom_bar(data = output_layer, aes(x = x, y = activation), stat = 'identity', position = 'dodge', fill = 'red', alpha = 0.3, width = 0.3) +
  annotate("text", x = output_x, y = rep(output_y, length(output_x)) - 0.3, label = c("Output #1", "Output #2", "Output #3")) +
  
  # Equation annotations
  annotate("text", x = 0.5, y = input_y + 1, label = "1", parse = TRUE) +
  annotate("text", x = 1.5, y = output_y - 1, label = "2 ", parse = TRUE) +
  
  # Input stimulus and output response
  annotate("text", x = mean(input_x), y = input_y + 1.3, label = "Input Stimulus") +
  annotate("text", x = mean(output_x), y = output_y - 1.3, label = "Output Response") +
  
  # Coordinate limits and axis labels
  coord_cartesian(xlim = c(0, 4), ylim = c(-1, 5)) +
  labs(x = "", y = "") +
  theme_void()

# Show the plot
print(p)

Alt ggplot 2

Code

library(ggplot2)
library(ggrepel)

neural_network_plot <- function(layers, layer_labels = NULL, node_labels = c('x', 'y')) {
data <- expand.grid(layer = seq_along(layers), node = 1:max(layers))
data <- data[data$node <= layers[data$layer], ]

links <- data.frame()
for (i in 1:(length(layers) - 1)) {
start <- subset(data, layer == i)
end <- subset(data, layer == i + 1)
links <- rbind(links, expand.grid(from = start$node, to = end$node, from_layer = start$layer, to_layer = end$layer))
}

p <- ggplot() +
geom_segment(data = links, aes(x = from_layer, xend = to_layer, y = from, yend = to), color = 'grey') +
geom_point(data = data, aes(x = layer, y = node), size = 5, color = 'orange') +
theme_minimal() +
theme(axis.text = element_blank(),
axis.ticks = element_blank(),
panel.grid = element_blank(),
axis.title = element_blank())

if (!is.null(layer_labels)) {
data$layer_label <- layer_labels[data$layer]
p <- p + geom_text_repel(data = data, aes(x = layer, y = max(layers) + 1, label = layer_label), size = 5, nudge_y = 1)
}

if (!is.null(node_labels)) {
data$node_label <- ifelse(data$layer == 1, paste0(node_labels[1], '', data$node),
ifelse(data$layer == length(layers), paste0(node_labels[2], '', data$node), ''))
p <- p + geom_text_repel(data = data, aes(x = layer, y = node, label = node_label), size = 3)
}

return(p)
}

#Example usage
layers <- c(8, 5, 3, 5, 8)
layer_labels <- c('Input Layer', '', 'Latent\nRepresentation', '', 'Output Layer')
neural_network_plot(layers, layer_labels)

Alt ggplot 3

Code

library(ggplot2)

data <- data.frame(
x = c(1, 2, 2, 3, 4, 4, 1, 2, 3, 4),
y = c(0, 1, -1, 0, 0, 0, 0, 0, 0, 0),
label = c("x_1", "w_1", "w_2", "Σ", "f", "y", "x_2", "w_3", "b", ""),
shape = c("circle", "square", "square", "circle", "square", "square", "circle", "square", "square", "square"),
text_y = c(0.15, 0.15, -0.15, 0, 0.3, 0.3, -0.15, -0.3, 0.15, 0)
)

data$shape <- factor(data$shape, levels = c("circle", "square"))

ggplot(data, aes(x = x, y = y)) +
geom_point(aes(shape = shape), size = 8) +
geom_text(aes(label = label, y = y + text_y), size = 5) +
scale_shape_manual(values = c("circle" = 19, "square" = 15)) +
geom_segment(aes(x = 2, xend = 3, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
geom_segment(aes(x = 2, xend = 3, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
geom_segment(aes(x = 3, xend = 2, y = 0.2, yend = 1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
geom_segment(aes(x = 3, xend = 2, y = -0.2, yend = -1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
geom_segment(aes(x = 3.5, xend = 4.5, y = 0, yend = 0)) +
geom_segment(aes(x = 4.5, xend = 5.5, y = 0, yend = 0)) +
annotate("text", x = 4, y = 0.5, label = "Activation\nFunction", hjust = 0.5) +
annotate("text", x = 5, y = 0, label = "Output", hjust = 0.5) +
annotate("text", x = 3, y = 1.1, label = "Bias\nb", hjust = 0.5) +
geom_rect(aes(xmin = 0.5, xmax = 1.5, ymin = -1.5, ymax = 1.5), color = "black", linetype = "dashed", fill = NA) +
annotate("text", x = 0.3, y = 0, label = "Inputs", angle = 90) +
theme_void() +
xlim(0, 6) +
theme(legend.position = "none")

Code

data <- data.frame(
    x = c(2, 3, 3, 4, 5, 5, 2, 3, 4, 5),
    y = c(0, 1, -1, 0, 0, 0, 0, 0, 0, 0),
    label = c("X", "w_{ji}", "(ai)", "(sum)", "O_j(X)", "m(X)", "X_i", "(e^{-gamma * (X-X[i])^2})", "(e^{-gamma * (Z-Y[j])^2})", "f_j(Z)"),
    shape = c("circle", "square", "circle", "circle", "circle", "circle", "circle", "circle", "circle", "circle"),
    text_y = c(0.15, 0.15, -0.15, 0, 0.3, 0.3, -0.15, -0.3, 0.15, 0)
)

data$shape <- factor(data$shape, levels = c("circle", "square"))

ggplot(data, aes(x = x, y = y)) +
    geom_point(aes(shape = shape), size = 8) +
    geom_text(aes(label = label, y = y + text_y), size = 4) +
    scale_shape_manual(values = c("circle" = 19, "square" = 15)) +
    geom_segment(aes(x = 3, xend = 4, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    geom_segment(aes(x = 3, xend = 4, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    geom_segment(aes(x = 4.5, xend = 5.5, y = 0, yend = 0)) +
    annotate("text", x = 5, y = 0.5, label = "Mean Output", hjust = 0.5) +
    annotate("text", x = 4, y = 1.1, label = "Normalized Input\nActivation", hjust = 0.5) +
    annotate("text", x = 4, y = -1.1, label = "Feedback\nActivation", hjust = 0.5) +
    geom_rect(aes(xmin = 1.5, xmax = 2.5, ymin = -1.5, ymax = 1.5), color = "black", linetype = "dashed", fill = NA) +
    annotate("text", x = 1.3, y = 0, label = "Input\nStimulus", angle = 90) +
    annotate("text", x = 3.5, y = -2, label = "Associative Learning Model - Activation and Learning", size = 5, hjust = 0.5) +
    theme_void() +
    xlim(0, 7) +
    theme(legend.position = "none")

Code

data <- data.frame(
    x = c(rep(2, 3), rep(4, 2), rep(6, 3)),
    y = c(-1, 0, 1, 0, 0, -1, 0, 1),
    label = c("e^{-γ(X-X_i)^2}", "Σ", "e^{-γ(Z-Y_j)^2}", "w_{ji}", "a_i(X)", "w_{ji}(t+1)", "O_j(X)", "Y_j"),
    shape = c("circle", "circle", "circle", "square", "square", "circle", "circle", "circle"),
    text_y = c(0.15, 0.15, 0.15, 0, 0, 0.15, 0.15, 0.15)
)

data$shape <- factor(data$shape, levels = c("circle", "square"))

ggplot(data, aes(x = x, y = y)) +
    geom_point(aes(shape = shape), size = 8) +
    geom_text(aes(label = label, y = y + text_y), size = 5) +
    scale_shape_manual(values = c("circle" = 19, "square" = 15)) +
    geom_segment(aes(x = 2, xend = 4, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    geom_segment(aes(x = 2, xend = 4, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    geom_segment(aes(x = 4, xend = 6, y = 0, yend = 1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    geom_segment(aes(x = 4, xend = 6, y = 0, yend = -1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) +
    annotate("text", x = 3, y = 1.5, label = "Input Layer\nGaussian Activation\na_i(X)=e^{-γ(X-X_i)^2}", hjust = 0.5) +
    annotate("text", x = 5, y = 1.5, label = "Output Layer\nO_j(X)=Σ w_{ji} a_i(X)\nm(X)=Σ Y_j P[Y_j | X]", hjust = 0.5) +
    annotate("text", x = 5, y = -1.5, label = "Learning\nw_{ji}(t+1)=w_{ji}(t)+α{f_j[Z(t)]-O_j[X(t)]}a_i[X(t)]\nf_j(Z)=e^{-γ(Z-Y_j)^2}", hjust = 0.5) +
    theme_void() +
    xlim(0, 8) +
    theme(legend.position = "none")

Code

# Load the required libraries
library(ggplot2)
library(ggpp)
library(tidyverse)
library(patchwork)

# Create the Gaussian curve plots for each input node
input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 1, sd = .4)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 2, sd = .4)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 3, sd = .4)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

# Create the main plot
main_plot <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("plot_npc", npcx = 0.2, npcy = 0.8, label = input_node_1) +
  annotate("plot_npc", npcx = 0.5, npcy = 0.8, label = input_node_2) +
  annotate("plot_npc", npcx = 0.8, npcy = 0.8, label = input_node_3) +
  annotate("text", x = 0.2, y = 0.7, label = "Input Node 1") +
  annotate("text", x = 0.5, y = 0.7, label = "Input Node 2") +
  annotate("text", x = 0.8, y = 0.7, label = "Input Node 3") +
  annotate("text", x = 0.5, y = 0.6, label = "Input Layer", fontface = "bold") +
  annotate("text", x = 0.5, y = 0.3, label = "Output Layer", fontface = "bold") +
  annotate("text", x = 0.5, y = 0.1, label = "ALM + EXAM Response", fontface = "bold")

# Print the main plot
print(main_plot)





# Create the Gaussian curve plots for each input node
input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 0, sd = 1)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 2.2, sd = 1)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 4.3, sd = .1)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0))

# Create the main plot
main_plot <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("plot_npc", npcx = 0.2, npcy = 0.8, label = input_node_1, width = 0.1, height = 0.1) +
  annotate("plot_npc", npcx = 0.5, npcy = 0.8, label = input_node_2, width = 0.1, height = 0.1) +
  annotate("plot_npc", npcx = 0.8, npcy = 0.8, label = input_node_3, width = 0.1, height = 0.1) +
  annotate("text", x = 0.2, y = 0.9, label = "Input Node 1") +
  annotate("text", x = 0.5, y = 0.9, label = "Input Node 2") +
  annotate("text", x = 0.8, y = 0.9, label = "Input Node 3") +
  annotate("text", x = 0.5, y = 0.6, label = "Input Layer", fontface = "bold") +
  annotate("rect", xmin = 0.1, xmax = 0.3, ymin = 0.4, ymax = 0.5, fill = "grey80") +
  annotate("rect", xmin = 0.4, xmax = 0.6, ymin = 0.4, ymax = 0.5, fill = "grey80") +
  annotate("rect", xmin = 0.7, xmax = 0.9, ymin = 0.4, ymax = 0.5, fill = "grey80") +
  annotate("text", x = 0.5, y = 0.3, label = "Output Layer", fontface = "bold") +
  annotate("text", x = 0.5, y = 0.1, label = "ALM + EXAM Response", fontface ="bold")
print(main_plot)


# Create the Gaussian curve plots for each input node
input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 0, sd = .41)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  ggtitle("Input Node 1")

input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 2, sd = .41)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  ggtitle("Input Node 2")

input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
  stat_function(fun = dnorm, args = list(mean = 3, sd = .41)) +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  ggtitle("Input Node 3")

# Create the output nodes
output_node_1 <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "blue") +
  ggtitle("Output Node 1")

output_node_2 <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "blue") +
  ggtitle("Output Node 2")

# Combine the input and output nodes using patchwork
input_layer <- input_node_1 + input_node_2 + input_node_3
output_layer <- output_node_1 + output_node_2

# Create the connection matrix
connection_matrix <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("segment", x = 0.2, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow()) +
  annotate("segment", x = 0.5, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow()) +
  annotate("segment", x = 0.8, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow())

# Combine the input layer, connection matrix, and output layer
main_plot <- input_layer / connection_matrix / output_layer

# Print the main plot
print(main_plot)






# Function to create an input node
create_input_node <- function(mean, position) {
  ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) +
    stat_function(fun = dnorm, args = list(mean = mean, sd = .3)) +
    theme_void() +
    theme(plot.margin = margin(0, 0, 0, 0)) +
    coord_cartesian(xlim = c(0, 3), ylim = c(0, 0.5)) + coord_flip()+
    labs(title = paste("Input Node", position)) +
    theme(plot.title = element_text(hjust = 0.5))
}

# Function to create an output node
create_output_node <- function(position) {
  ggplot() +
    theme_void() +
    theme(plot.margin = margin(0, 0, 0, 0)) +
    annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "grey80") +
    labs(title = paste("Output Node", position)) +
    theme(plot.title = element_text(hjust = 0.5))
}

create_input_layer <- function(n) {
  input_plots <- map(1:n, ~create_gaussian_plot(.x))
  input_layer <- wrap_plots(input_plots, ncol = 1) + coord_flip()
  return(input_layer)
}



# Create the input nodes
input_nodes <- map(1:3, ~create_input_node(.x, .x))

# Create the output nodes
output_nodes <- map(1:4, ~create_output_node(.x))

## Create the connection matrix
connection_matrix <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("segment", x = 1, xend = 2, y = rep(1:3, each = 4), yend = rep(1:4, times = 3), arrow = arrow())
# Combine the plots
input_layer <- wrap_plots(input_nodes,byrow=FALSE,ncol=1)
output_layer <- wrap_plots(output_nodes, ncol = 1)
main_plot <- input_layer + connection_matrix + output_layer + plot_layout(ncol = 3)

# Print the main plot
print(main_plot)





combined_plot <- input_layer + connection_matrix + output_layer +
  plot_layout(ncol = 3)

print(combined_plot)



print(input_layer)

Code

# Load the required libraries
library(ggplot2)
library(ggpp)
library(patchwork)
library(tidyverse)

# Function to create the input layer
create_input_layer <- function(n_nodes, stimulus = 0, gamma = 5) {
  # Create a data frame with the x values and node means
  df <- expand.grid(x = seq(-3, 3, length.out = 100),
                    mean = seq(-2, 1.5, length.out = n_nodes)) %>% 
    mutate(noisyMean=map(mean,~dnorm(.x,mean,sd=.5)))
  
  df <- df %>%
    mutate(distance = abs(x - stimulus),
           noise = rnorm(n(), sd = 0.05),
           max_activation = exp(-gamma * distance^2) + noise,
           y = max_activation * exp(-gamma * (x - mean)^2))
  
  # Create the plot
  ggplot(df, aes(x, y, group = mean)) +
    geom_line() +
    geom_smooth()+
    coord_cartesian(xlim = c(-2, 2), ylim = c(0, 1)) +
    labs(title = "Input Layer") +
    theme(plot.title = element_text(hjust = 0.5))
}

# Function to create an output node
create_output_node <- function(position) {
  ggplot() +
    theme_void() +
    theme(plot.margin = margin(0, 0, 0, 0)) +
    annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "grey80") +
    labs(title = paste("Output Node", position)) +
    theme(plot.title = element_text(hjust = 0.5))
}

# Create the input layer
input_layer <- create_input_layer(3)

# Create the output nodes
output_nodes <- map(1:4, ~create_output_node(.x))

# Create the connection matrix
connection_matrix <- ggplot() +
  theme_void() +
  theme(plot.margin = margin(0, 0, 0, 0)) +
  annotate("segment", x = 1, xend = 2, y = 1:3, yend = rep(1:4, each = 3), arrow = arrow())

# Combine the plots
output_layer <- wrap_plots(output_nodes, ncol = 1)
main_plot <- input_layer + connection_matrix + output_layer + plot_layout(ncol = 3)

# Print the main plot
print(main_plot)

Code

# Load the visNetwork package
library(visNetwork)

# Define the nodes
nodes <- data.frame(
  id = c("InputLayer", "I1", "I2", "I3", "OutputLayer", "O1", "O2", "O3", "O4", "ALM", "EXAM"),
  label = c("Input Layer", "Input Node 1", "Input Node 2", "Input Node 3", "Output Layer", "Output Node 1", "Output Node 2", "Output Node 3", "Output Node 4", "ALM Response", "EXAM Response"),
  shape = c("box", "image", "image", "image", "box", "circle", "circle", "circle", "circle", "box", "box"),
  image = c(NA, "gaussian_curve_input_node_1.png", "gaussian_curve_input_node_2.png", "gaussian_curve_input_node_3.png", NA, NA, NA, NA, NA, NA, NA)
)

# Define the edges
edges <- data.frame(
  from = c("InputLayer", "I1", "I1", "I1", "I1", "I2", "I2", "I2", "I2", "I3", "I3", "I3", "I3", "OutputLayer", "OutputLayer", "ALM", "EXAM"),
  to = c("I1", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "ALM", "EXAM", "EXAM", "ALM"),
  label = c("", "w1", "w2", "w3", "w4", "w5", "w6", "w7", "w8", "w9", "w10", "w11", "w12", "", "", "Pure ALM Model", "ALM with EXAM Response Component")
)

# Create the network diagram
visNetwork(nodes, edges, width = "100%") %>%
  visNodes(shapeProperties = list(useBorderWithImage = TRUE)) %>%
  visLayout(randomSeed = 2)





# Define the nodes
nodes <- data.frame(
  id = c("InputLayer", "I1", "I2", "I3", "OutputLayer", "O1", "O2", "O3", "O4", "ALM", "EXAM"),
  label = c("Input Layer", "<img src='gaussian_curve_input_node_1.png' />", "<img src='gaussian_curve_input_node_2.png' />", "<img src='gaussian_curve_input_node_3.png' />", "Output Layer", "Output Node 1", "Output Node 2", "Output Node 3", "Output Node 4", "ALM Response", "EXAM Response"),
  shape = c("box", "circle", "circle", "circle", "box", "circle", "circle", "circle", "circle", "box", "box")
)

# Define the edges
edges <- data.frame(
  from = c("InputLayer", "I1", "I1", "I1", "I1", "I2", "I2", "I2", "I2", "I3", "I3", "I3", "I3", "OutputLayer", "OutputLayer", "ALM", "EXAM"),
  to = c("I1", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "ALM", "EXAM", "EXAM", "ALM"),
  label = c(NA, "w1", "w2", "w3", "w4", "w5", "w6", "w7", "w8", "w9", "w10", "w11", "w12", NA, NA, "Pure ALM Model", "ALM with EXAM Response Component")
)

# Create the network diagram
visNetwork(nodes, edges, width = "100%") %>%
  visNodes(shapeProperties = list(useBorderWithImage = TRUE)) %>%
  visLayout(randomSeed = 2)

Code

# Load the DiagrammeR package
library(DiagrammeR)
# Define the graph
grViz("
  digraph ALM_EXAM {

    # Graph attributes
    graph [overlap = true, fontsize = 10]

    # Node definitions
    node [shape = box, fontname = Helvetica]
    InputLayer [label = 'Input Layer']
    OutputLayer [label = 'Output Layer']
    ALM [label = 'ALM Response']
    EXAM [label = 'EXAM Response']

    node [shape = circle, fixedsize = true, width = 0.9]
    I1 [label = 'Input Node 1']
    I2 [label = 'Input Node 2']
    I3 [label = 'Input Node 3']
    O1 [label = 'Output Node 1']
    O2 [label = 'Output Node 2']
    O3 [label = 'Output Node 3']
    O4 [label = 'Output Node 4']

    # Edge definitions
    InputLayer -> I1
    InputLayer -> I2
    InputLayer -> I3
    I1 -> O1 [label = 'w1']
    I1 -> O2 [label = 'w2']
    I1 -> O3 [label = 'w3']
    I1 -> O4 [label = 'w4']
    I2 -> O1 [label = 'w5']
    I2 -> O2 [label = 'w6']
    I2 -> O3 [label = 'w7']
    I2 -> O4 [label = 'w8']
    I3 -> O1 [label = 'w9']
    I3 -> O2 [label = 'w10']
    I3 -> O3 [label = 'w11']
    I3 -> O4 [label = 'w12']
    OutputLayer -> O1
    OutputLayer -> O2
    OutputLayer -> O3
    OutputLayer -> O4
    OutputLayer -> ALM
    OutputLayer -> EXAM
    ALM -> EXAM [label = 'Pure ALM Model']
    EXAM -> ALM [label = 'ALM with EXAM Response Component']
  }
")

Code

# Load the required packages
library(ggplot2)
library(png)

# Define the Gaussian function
gaussian <- function(x, mean, sd) {
  1/(sd*sqrt(2*pi)) * exp(-1/2 * ((x - mean)/sd)^2)
}

# Define the central values for the input nodes
central_values <- c(1, 2, 3)

# Define the input stimulus value
input_stimulus <- 2

# Generate a Gaussian curve image for each input node
for (i in seq_along(central_values)) {
  # Define the mean and standard deviation for the Gaussian curve
  mean <- central_values[i]
  sd <- 1
  # Generate the x values
  x <- seq(mean - 3*sd, mean + 3*sd, length.out = 100)
  # Generate the y values
  y <- gaussian(x, mean, sd)
  # Create the plot
  p <- ggplot(data.frame(x, y), aes(x, y)) +
    geom_line() +
    theme_minimal() +
    labs(x = "X", y = "Activation", title = paste("Input Node", i))
  # Save the plot as a PNG image
  ggsave(paste0("gaussian_curve_input_node_", i, ".png"), plot = p, width = 4, height = 3)
}

Code

grViz("digraph causal {
                        # Nodes
                        node [imagescale=true,shape = reactangle, fontname = Arial, style = filled]
                        iv   [label = 'TRT', fillcolor = '#7FC97F']
                        me   [label = 'Mediator', shape = ellipse]
                        dv   [label = 'DLQI', fillcolor = '#7FC97F']
                        
                        # Edges
                        edge [color = black, arrowhead = normal]
                        rankdir = LR
                        iv -> me
                        iv -> dv [label = 'DIRECT', fontcolor = '#7FC97F', color = '#7FC97F']
                        me -> dv
                        # Graph
                        graph [overlap = true, fontsize = 10]
                      }")


grViz("digraph causal {
                        # Nodes
                        node [shape = reactangle, fontname = Arial, style = filled]
                        iv   [image='gaussian_curve_input_node_1.png',label = '', fillcolor = '#7FC97F']
                        me   [label = 'Mediator', shape = ellipse]
                        dv   [label = 'DLQI', fillcolor = '#7FC97F']
                        
                        # Edges
                        edge [color = black, arrowhead = normal]
                        rankdir = LR
                        iv -> me
                        iv -> dv [label = 'DIRECT', fontcolor = '#7FC97F', color = '#7FC97F']
                        me -> dv
                        # Graph
                        graph [overlap = true, fontsize = 10]
                      }")

Code

# grViz('digraph structs {
#     node [shape=plaintext];
# 
#     struct1 [label=<<TABLE>
#       <TR><TD><IMG SRC="gaussian_curve_input_node_1.png"/></TD></TR>
#       <TR><TD>caption</TD></TR>
#     </TABLE>>];
# }')


svgDataUrl = "data:image/svg+xml;base64,"
grViz('
digraph {
  node[imagescale=true, shape=circle, width=4, height=4,fontname="sans-serif",penwidth=0]
  a[label=<<FONT POINT-SIZE="10">Transaction</FONT><BR/><FONT POINT-SIZE="36"><B>TRNA</B></FONT>>,image="${svgDataUrl}"]
  b[label=<<FONT POINT-SIZE="10">Transaction</FONT><BR/><FONT POINT-SIZE="18"><B>TRNB</B></FONT>>,image="${svgDataUrl}"]
  c[label=<<FONT POINT-SIZE="10">Transaction</FONT><BR/><FONT POINT-SIZE="24"><B>TRNC</B></FONT>>,image="${svgDataUrl}"]
  a -> b
  a -> c
}')

grViz('digraph { a[image=\"gaussian_curve_input_node_1\"]; }", {
  images: [{ href: "gaussian_curve_input_node_1.png", width: "400px", height: "300px" }]
}');

grViz('
digraph g{
  node[imagescale=true, shape=circle, width=14, height=14,fontname="sans-serif",penwidth=0]
  I1[image="gaussian_curve_input_node_1.png", label=""];
}')

grViz('digraph {
    ratio="fill";
    size="10,10!";
    margin="0,0";

    node [shape=plain];
    root [label=<<TABLE border="0"><TR><TD><IMG SRC="gaussian_curve_input_node_1"/></TD></TR>
                                   <TR><TD>text under</TD></TR></TABLE>>];
}')

Code

grViz("
  digraph ALM_EXAM {

    # Graph attributes
    graph [overlap = true, fontsize = 10, rankdir = LR]

    # Node definitions
    node [shape = box, fontname = Helvetica]
    InputLayer [label = 'Input Layer']
    OutputLayer [label = 'Output Layer']
    ALM [label = 'ALM Response']
    EXAM [label = 'EXAM Response']

    node [shape = none, label = '']
    I1 [image = 'gaussian_curve_input_node_1.png', label='']
    I2 [image = 'gaussian_curve_input_node_2.png', label='']
    I3 [image = 'gaussian_curve_input_node_3.png', label='']

    node [shape = circle, fixedsize = true, width = 0.9]
    O1 [label = 'Output Node 1']
    O2 [label = 'Output Node 2']
    O3 [label = 'Output Node 3']
    O4 [label = 'Output Node 4']

    # Edge definitions
    InputLayer -> I1
    InputLayer -> I2
    InputLayer -> I3
    I1 -> O1 [label = 'w1']
    I1 -> O2 [label = 'w2']
    I1 -> O3 [label = 'w3']
    I1 -> O4 [label = 'w4']
    I2 -> O1 [label = 'w5']
    I2 -> O2 [label = 'w6']
    I2 -> O3 [label = 'w7']
    I2 -> O4 [label = 'w8']
    I3 -> O1 [label = 'w9']
    I3 -> O2 [label = 'w10']
    I3 -> O3 [label = 'w11']
    I3 -> O4 [label = 'w12']
    OutputLayer -> O1
    OutputLayer -> O2
    OutputLayer -> O3
    OutputLayer -> O4
    OutputLayer -> ALM
    OutputLayer -> EXAM
    ALM -> EXAM [label = 'Pure ALM Model']
    EXAM -> ALM [label = 'ALM with EXAM Response Component']
  }
")

Code

DiagrammeR::grViz("digraph {

graph [layout = dot, rankdir = LR]

# define the global styles of the nodes. We can override these in box if we wish
node [shape = rectangle, style = filled, fillcolor = Linen]

data1 [label = 'Dataset 1', shape = folder, fillcolor = Beige]
data2 [label = 'Dataset 2', shape = folder, fillcolor = Beige]
process [label =  'Process \n Data']
statistical [label = 'Statistical \n Analysis']
results [label= 'Results']

# edge definitions with the node IDs
{data1 data2}  -> process -> statistical -> results
}")

graph LR
  subgraph InputLayer
    I1["Input Node 1"]
    I2["Input Node 2"]
    I3["Input Node 3"]
  end
  subgraph OutputLayer
    O1["Output Node 1"]
    O2["Output Node 2"]
    O3["Output Node 3"]
    O4["Output Node 4"]
  end
  I1 -->|"w1"| O1
  I1 -->|"w2"| O2
  I1 -->|"w3"| O3
  I1 -->|"w4"| O4
  I2 -->|"w5"| O1
  I2 -->|"w6"| O2
  I2 -->|"w7"| O3
  I2 -->|"w8"| O4
  I3 -->|"w9"| O1
  I3 -->|"w10"| O2
  I3 -->|"w11"| O3
  I3 -->|"w12"| O4
  style InputLayer fill:#99cc99,stroke:#333,stroke-width:2px
  style OutputLayer fill:#cc99cc,stroke:#333,stroke-width:2px
  IS["Input Stimulus (X)"] --> InputLayer
  InputLayer -->|"Gaussian Activation (a_i)"| OutputLayer
  OutputLayer --> ALM["ALM Response (m(X))"]
  OutputLayer --> EXAM["EXAM Response (E[Y | X_i])"]
  ALM -->|"Pure ALM Model"| EXAM
  EXAM -->|"ALM with EXAM Response Component"| ALM
  linkStyle 0 stroke:#2ecd71,stroke-width:2px;
  linkStyle 1 stroke:#2ecd71,stroke-width:2px;
  linkStyle 2 stroke:#2ecd71,stroke-width:2px;
  linkStyle 3 stroke:#2ecd71,stroke-width:2px;
  linkStyle 4 stroke:#2ecd71,stroke-width:2px;
  linkStyle 5 stroke:#2ecd71,stroke-width:2px;
  linkStyle 6 stroke:#2ecd71,stroke-width:2px;
  linkStyle 7 stroke:#2ecd71,stroke-width:2px;
  linkStyle 8 stroke:#2ecd71,stroke-width:2px;
  linkStyle 9 stroke:#2ecd71,stroke-width:2px;
  linkStyle 10 stroke:#2ecd71,stroke-width:2px;
  linkStyle 11 stroke:#2ecd71,stroke-width:2px;
  linkStyle 12 stroke:#2ecd71,stroke-width:2px;
  linkStyle 13 stroke:#2ecd71,stroke-width:2px;
  linkStyle 14 stroke:#2ecd71,stroke-width:2px;
  linkStyle 15 stroke:#2ecd71,stroke-width:2px;
  linkStyle 16 stroke:#2ecd71,stroke-width:2px;
  linkStyle 17 stroke:#2ecd71,stroke-width:2px;
  linkStyle 0 stroke:#2ecd71,stroke-width:2px;
  linkStyle 1 stroke:#2ecd71,stroke-width:2px;
  linkStyle 2 stroke:#2ecd71,stroke-width:2px;
  linkStyle 3 stroke:#2ecd71,stroke-width:2px;
  linkStyle 4 stroke:#2ecd71,stroke-width:2px;
  linkStyle 5 stroke:#2ecd71,stroke-width:2px;
  linkStyle 6 stroke:#2ecd71,stroke-width:2px;
  linkStyle 7 stroke:#2ecd71,stroke-width:2px;
  linkStyle 8 stroke:#2ecd71,stroke-width:2px;
  linkStyle 9 stroke:#2ecd71,stroke-width:2px;
  linkStyle 10 stroke:#2ecd71,stroke-width:2px;
  linkStyle 11 stroke:#2ecd71,stroke-width:2px;
  linkStyle 12 stroke:#2ecd71,stroke-width:2px;
  linkStyle 13 stroke:#2ecd71,stroke-width:2px;
  linkStyle 14 stroke:#2ecd71,stroke-width:2px;
  linkStyle 15 stroke:#2ecd71,stroke-width:2px;
  linkStyle 16 stroke:#2ecd71,stroke-width:2px;
  linkStyle 17 stroke:#2ecd71,stroke-width:2px;

P5 sketching

Code

P5 = require("p5")
function* createSketch(sketch) {
  const element = DOM.element('div');
  yield element;
  const instance = new P5(sketch, element, true);
  try {
    while (true) {
      yield element;
    }
  } finally {
    instance.remove();
  }
}
createSketch(s => {
  
    s.setup = function() {
      s.createCanvas(746, 300);
      s.textFont('Courgette');
      s.textStyle(s.BOLD);
      s.textAlign(s.CENTER, s.CENTER)

      s.button = s.createButton('clear');
      s.button.mousePressed(s.clearCanvas);
        s.text('Click and drag to draw', s.width/2, s.height/10);

    };
    s.draw = function() {
    if (s.mouseIsPressed) {
    s.fill(0);
    s.ellipse(s.mouseX, s.mouseY, 10, 10);
    } else {
   //s.fill(255);
    }
  // add text input

    };

  // add button to clear canvas
  s.clearCanvas = function() {
    s.clear();
  };
  // add text
  // add slider
  }
)

--- title: Model Visualization date: last-modified categories: [Visualization, R, OJS] page-layout: full code-fold: true code-tools: true execute: warning: false --- ```{r} #lapply(c('tidyverse','data.table','igraph','ggraph','kableExtra'),library,character.only=TRUE)) pacman::p_load(tidyverse,data.table,igraph,ggraph,kableExtra,DiagrammeR,png, plantuml) ``` ### plantuml visualizations ```{r} uml <- '@startuml skinparam ranksep 20 skinparam dpi 125 skinparam packageTitleAlignment left rectangle "Input Layer" { (Input1) (Input2) } rectangle "Output Layer" { (Output1) (Output2) (Output3) } rectangle "ALM (Associative Learning Model)" as ALM_model (Input1) ..> (Output1) : w_{11} (Input1) ..> (Output2) : w_{12} (Input1) ..> (Output3) : w_{13} (Input2) ..> (Output1) : w_{21} (Input2) ..> (Output2) : w_{22} (Input2) ..> (Output3) : w_{23} ALM_model ==> (Input Layer) (Input Layer) --> (Output Layer) @enduml ' uml <- plantuml(uml) plot(x=uml) ``` ```{r} alm_model_with_equations <- ' @startuml skinparam ranksep 20 skinparam dpi 125 skinparam packageTitleAlignment left rectangle "Input Layer" { (Input1) (Input2) } rectangle "Output Layer" { (Output1) (Output2) (Output3) } (Input1) ..> (Output1) : w_{11} (Input1) ..> (Output2) : w_{12} (Input1) ..> (Output3) : w_{13} (Input2) ..> (Output1) : w_{21} (Input2) ..> (Output2) : w_{22} (Input2) ..> (Output3) : w_{23} rectangle "ALM (Associative Learning Model)" as ALM_model ALM_model ==> (Input Layer) (Input Layer) --> (Output Layer) note right of (Input Layer) <math>a_i(X)=e^{-\\gamma \\cdot(X-X_i)^2}</math> <math>Normalized: \\frac{a_i(X)}{\\sum a_i(X)}</math> end note note right of (Output Layer) blank end note @enduml ' # Pass the PlantUML code to plantuml function alm_model_with_equations <- plantuml(alm_model_with_equations) plot(x=alm_model_with_equations) ``` ## ggplot model visualization ```{r} nInput=6 nOutput=10 inNodes <- seq(1,nInput,1) %>% as.integer() outNodes <- seq(300,1000,length.out=nOutput)%>% as.integer() weight.mat <<- matrix(0.001,nrow=nOutput,ncol=nInput) # weights initialized to 0 (as in Delosh 1997) stim <- 1.5 c=.1 inAct <- round(exp(-c*((inNodes-stim)^2)),2) inActLab <- paste0("x",inNodes,"=",inAct) outAct <- weight.mat %*% inAct output.probability <<- outAct/sum(outAct) outLab=paste0("y",outNodes,"=",round(output.probability,2)) mean.response <<- round(sum(outNodes * output.probability),0) resp <- mean.response inFlow <- tibble(expand.grid(from=stim,to=inActLab)) %>% mutate_all(as.character) outFlow <- tibble(expand.grid(from=outLab,to=mean.response)) %>% mutate_all(as.character) gd <- tibble(expand.grid(from=inActLab,to=outLab)) %>% mutate_all(as.character) %>% rbind(inFlow,.) %>% rbind(.,outFlow) xInc <- .3 yInc=.5 g = graph_from_data_frame(gd,directed=TRUE) coords2=layout_as_tree(g) colnames(coords2)=c("y","x") odf <- as_tibble(coords2) %>% mutate(label=vertex_attr(g,"name"), type=c("stim",rep("Input",nInput),rep("Output",nOutput),"Resp"), x=x*-1) %>% mutate(y=ifelse(type=="Resp",0,y),xmin=x-xInc,xmax=x+xInc,ymin=y-yInc,ymax=y+yInc) plot_edges = gd %>% mutate(id=row_number()) %>% pivot_longer(cols=c("from","to"),names_to="s_e",values_to=("label")) %>% mutate(label=as.character(label)) %>% group_by(id) %>% mutate(weight=sqrt(rnorm(1,mean=0,sd=10)^2)/10) %>% left_join(odf,by="label") %>% mutate(xmin=xmin+.02,xmax=xmax-.02) ggplot() + geom_rect(data = odf, mapping = aes(xmin = xmin, ymin = ymin, xmax = xmax, ymax = ymax, fill = type, colour = type),alpha = 0.01) + geom_text(data=odf,aes(x=x,y=y,label=label,size=3)) + geom_path(data=plot_edges,mapping=aes(x=x,y=y,group=id,alpha=weight)) + # geom_rect(aes(xmin=-1.05,xmax=-.95,ymin=-10,ymax=5),color="red",alpha=.1)+ # geom_rect(aes(xmin=-0.05,xmax=.05,ymin=-10,ymax=5),color="blue",alpha=.1) + theme_void() ``` ## ggraph method ```{r } inNodes <- seq(1,6,1) %>% as.integer() outNodes <- seq(300,1000,50)%>% as.integer() da <- data.frame(expand.grid(inNodes,outNodes)) %>% magrittr::set_colnames(c("input","output")) da <- da %>% mutate_all(as.character) m = graph_from_data_frame(da, directed = TRUE) coords = layout_with_sugiyama(m) colnames(coords$layout) = c("y", "x") coords$layout=coords$layout[,c("x","y")] plot(m,layout=coords) ggraph(m,layout=coords$layout)+ geom_edge_link0(width=0.2,colour="grey")+ geom_node_point(col="white",size=6)+scale_x_reverse()+ geom_node_text(aes(label=name)) + # draw rectangle that covers input layer at x=1, min y is min of coords$y and max y is max of coords$y annotate("rect",xmin=0,xmax=.1,ymin=min(coords$layout[,2]),ymax=max(coords$layout[,2]),fill="grey",alpha=0.7) geom_rect(xmin=0,xmax=1.1,ymin=min(coords$layout[,2]),ymax=max(coords$layout[,2]),fill="grey",alpha=0.7) ``` ```{r} library(tidyverse) library(ggforce) theme_set(theme_grey() + theme_void() + theme(plot.margin = margin(0, 5.5, 0, 5.5))) # Parameters input_x <- c(1, 2) output_x <- c(1.5, 2.5, 3.5) input_y <- 3 output_y <- 1 activation_levels <- c(0.2, 0.7, 0.5) # Input Layer input_layer <- tibble(x = input_x, y = rep(input_y, length(input_x))) # Output Layer output_layer <- tibble(x = output_x, y = rep(output_y, length(output_x)), activation = activation_levels) # Gaussian Activation gaussian_activation <- tibble( x = seq(from = 0, to = 3, by = 0.01), y = exp(-2 * (x - 1)^2) + 2.5, x2 = x, y2 = exp(-2 * (x - 2)^2) + 2.5 ) # Plot p <- ggplot() + # Input nodes geom_point(data = input_layer, aes(x = x, y = y), size = 4, color = 'green') + annotate("text", x = input_x, y = rep(input_y, length(input_x)) + 0.3, label = c("Input #1", "Input #2")) + # Gaussian Activations geom_line(data = gaussian_activation, aes(x = x, y = y), color = 'blue') + geom_line(data = gaussian_activation, aes(x = x2, y = y2), color = 'blue') + # Connections geom_segment(data = expand.grid(input_x, output_x), aes(x = Var1, xend = Var2, y = input_y, yend = output_y), arrow = arrow(type = 'closed', length = unit(0.2, 'inches'))) + # Output nodes geom_point(data = output_layer, aes(x = x, y = y), size = 4, color = 'red') + geom_bar(data = output_layer, aes(x = x, y = activation), stat = 'identity', position = 'dodge', fill = 'red', alpha = 0.3, width = 0.3) + annotate("text", x = output_x, y = rep(output_y, length(output_x)) - 0.3, label = c("Output #1", "Output #2", "Output #3")) + # Equation annotations annotate("text", x = 0.5, y = input_y + 1, label = "1", parse = TRUE) + annotate("text", x = 1.5, y = output_y - 1, label = "2 ", parse = TRUE) + # Input stimulus and output response annotate("text", x = mean(input_x), y = input_y + 1.3, label = "Input Stimulus") + annotate("text", x = mean(output_x), y = output_y - 1.3, label = "Output Response") + # Coordinate limits and axis labels coord_cartesian(xlim = c(0, 4), ylim = c(-1, 5)) + labs(x = "", y = "") + theme_void() # Show the plot print(p) ``` ### Alt ggplot 2 ```{r} library(ggplot2) library(ggrepel) neural_network_plot <- function(layers, layer_labels = NULL, node_labels = c('x', 'y')) { data <- expand.grid(layer = seq_along(layers), node = 1:max(layers)) data <- data[data$node <= layers[data$layer], ] links <- data.frame() for (i in 1:(length(layers) - 1)) { start <- subset(data, layer == i) end <- subset(data, layer == i + 1) links <- rbind(links, expand.grid(from = start$node, to = end$node, from_layer = start$layer, to_layer = end$layer)) } p <- ggplot() + geom_segment(data = links, aes(x = from_layer, xend = to_layer, y = from, yend = to), color = 'grey') + geom_point(data = data, aes(x = layer, y = node), size = 5, color = 'orange') + theme_minimal() + theme(axis.text = element_blank(), axis.ticks = element_blank(), panel.grid = element_blank(), axis.title = element_blank()) if (!is.null(layer_labels)) { data$layer_label <- layer_labels[data$layer] p <- p + geom_text_repel(data = data, aes(x = layer, y = max(layers) + 1, label = layer_label), size = 5, nudge_y = 1) } if (!is.null(node_labels)) { data$node_label <- ifelse(data$layer == 1, paste0(node_labels[1], '', data$node), ifelse(data$layer == length(layers), paste0(node_labels[2], '', data$node), '')) p <- p + geom_text_repel(data = data, aes(x = layer, y = node, label = node_label), size = 3) } return(p) } #Example usage layers <- c(8, 5, 3, 5, 8) layer_labels <- c('Input Layer', '', 'Latent\nRepresentation', '', 'Output Layer') neural_network_plot(layers, layer_labels) ``` ### Alt ggplot 3 ```{r} library(ggplot2) data <- data.frame( x = c(1, 2, 2, 3, 4, 4, 1, 2, 3, 4), y = c(0, 1, -1, 0, 0, 0, 0, 0, 0, 0), label = c("x_1", "w_1", "w_2", "Σ", "f", "y", "x_2", "w_3", "b", ""), shape = c("circle", "square", "square", "circle", "square", "square", "circle", "square", "square", "square"), text_y = c(0.15, 0.15, -0.15, 0, 0.3, 0.3, -0.15, -0.3, 0.15, 0) ) data$shape <- factor(data$shape, levels = c("circle", "square")) ggplot(data, aes(x = x, y = y)) + geom_point(aes(shape = shape), size = 8) + geom_text(aes(label = label, y = y + text_y), size = 5) + scale_shape_manual(values = c("circle" = 19, "square" = 15)) + geom_segment(aes(x = 2, xend = 3, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 2, xend = 3, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 3, xend = 2, y = 0.2, yend = 1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 3, xend = 2, y = -0.2, yend = -1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 3.5, xend = 4.5, y = 0, yend = 0)) + geom_segment(aes(x = 4.5, xend = 5.5, y = 0, yend = 0)) + annotate("text", x = 4, y = 0.5, label = "Activation\nFunction", hjust = 0.5) + annotate("text", x = 5, y = 0, label = "Output", hjust = 0.5) + annotate("text", x = 3, y = 1.1, label = "Bias\nb", hjust = 0.5) + geom_rect(aes(xmin = 0.5, xmax = 1.5, ymin = -1.5, ymax = 1.5), color = "black", linetype = "dashed", fill = NA) + annotate("text", x = 0.3, y = 0, label = "Inputs", angle = 90) + theme_void() + xlim(0, 6) + theme(legend.position = "none") data <- data.frame( x = c(2, 3, 3, 4, 5, 5, 2, 3, 4, 5), y = c(0, 1, -1, 0, 0, 0, 0, 0, 0, 0), label = c("X", "w_{ji}", "(ai)", "(sum)", "O_j(X)", "m(X)", "X_i", "(e^{-gamma * (X-X[i])^2})", "(e^{-gamma * (Z-Y[j])^2})", "f_j(Z)"), shape = c("circle", "square", "circle", "circle", "circle", "circle", "circle", "circle", "circle", "circle"), text_y = c(0.15, 0.15, -0.15, 0, 0.3, 0.3, -0.15, -0.3, 0.15, 0) ) data$shape <- factor(data$shape, levels = c("circle", "square")) ggplot(data, aes(x = x, y = y)) + geom_point(aes(shape = shape), size = 8) + geom_text(aes(label = label, y = y + text_y), size = 4) + scale_shape_manual(values = c("circle" = 19, "square" = 15)) + geom_segment(aes(x = 3, xend = 4, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 3, xend = 4, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 4.5, xend = 5.5, y = 0, yend = 0)) + annotate("text", x = 5, y = 0.5, label = "Mean Output", hjust = 0.5) + annotate("text", x = 4, y = 1.1, label = "Normalized Input\nActivation", hjust = 0.5) + annotate("text", x = 4, y = -1.1, label = "Feedback\nActivation", hjust = 0.5) + geom_rect(aes(xmin = 1.5, xmax = 2.5, ymin = -1.5, ymax = 1.5), color = "black", linetype = "dashed", fill = NA) + annotate("text", x = 1.3, y = 0, label = "Input\nStimulus", angle = 90) + annotate("text", x = 3.5, y = -2, label = "Associative Learning Model - Activation and Learning", size = 5, hjust = 0.5) + theme_void() + xlim(0, 7) + theme(legend.position = "none") data <- data.frame( x = c(rep(2, 3), rep(4, 2), rep(6, 3)), y = c(-1, 0, 1, 0, 0, -1, 0, 1), label = c("e^{-γ(X-X_i)^2}", "Σ", "e^{-γ(Z-Y_j)^2}", "w_{ji}", "a_i(X)", "w_{ji}(t+1)", "O_j(X)", "Y_j"), shape = c("circle", "circle", "circle", "square", "square", "circle", "circle", "circle"), text_y = c(0.15, 0.15, 0.15, 0, 0, 0.15, 0.15, 0.15) ) data$shape <- factor(data$shape, levels = c("circle", "square")) ggplot(data, aes(x = x, y = y)) + geom_point(aes(shape = shape), size = 8) + geom_text(aes(label = label, y = y + text_y), size = 5) + scale_shape_manual(values = c("circle" = 19, "square" = 15)) + geom_segment(aes(x = 2, xend = 4, y = 1, yend = 0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 2, xend = 4, y = -1, yend = -0.2), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 4, xend = 6, y = 0, yend = 1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + geom_segment(aes(x = 4, xend = 6, y = 0, yend = -1), arrow = arrow(type = "closed", length = unit(0.2, "inches"))) + annotate("text", x = 3, y = 1.5, label = "Input Layer\nGaussian Activation\na_i(X)=e^{-γ(X-X_i)^2}", hjust = 0.5) + annotate("text", x = 5, y = 1.5, label = "Output Layer\nO_j(X)=Σ w_{ji} a_i(X)\nm(X)=Σ Y_j P[Y_j | X]", hjust = 0.5) + annotate("text", x = 5, y = -1.5, label = "Learning\nw_{ji}(t+1)=w_{ji}(t)+α{f_j[Z(t)]-O_j[X(t)]}a_i[X(t)]\nf_j(Z)=e^{-γ(Z-Y_j)^2}", hjust = 0.5) + theme_void() + xlim(0, 8) + theme(legend.position = "none") ``` ```{r fig.width=11, fig.height=12} #| eval: false # Load the required libraries library(ggplot2) library(ggpp) library(tidyverse) library(patchwork) # Create the Gaussian curve plots for each input node input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 1, sd = .4)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 2, sd = .4)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 3, sd = .4)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) # Create the main plot main_plot <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("plot_npc", npcx = 0.2, npcy = 0.8, label = input_node_1) + annotate("plot_npc", npcx = 0.5, npcy = 0.8, label = input_node_2) + annotate("plot_npc", npcx = 0.8, npcy = 0.8, label = input_node_3) + annotate("text", x = 0.2, y = 0.7, label = "Input Node 1") + annotate("text", x = 0.5, y = 0.7, label = "Input Node 2") + annotate("text", x = 0.8, y = 0.7, label = "Input Node 3") + annotate("text", x = 0.5, y = 0.6, label = "Input Layer", fontface = "bold") + annotate("text", x = 0.5, y = 0.3, label = "Output Layer", fontface = "bold") + annotate("text", x = 0.5, y = 0.1, label = "ALM + EXAM Response", fontface = "bold") # Print the main plot print(main_plot) # Create the Gaussian curve plots for each input node input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 0, sd = 1)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 2.2, sd = 1)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 4.3, sd = .1)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) # Create the main plot main_plot <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("plot_npc", npcx = 0.2, npcy = 0.8, label = input_node_1, width = 0.1, height = 0.1) + annotate("plot_npc", npcx = 0.5, npcy = 0.8, label = input_node_2, width = 0.1, height = 0.1) + annotate("plot_npc", npcx = 0.8, npcy = 0.8, label = input_node_3, width = 0.1, height = 0.1) + annotate("text", x = 0.2, y = 0.9, label = "Input Node 1") + annotate("text", x = 0.5, y = 0.9, label = "Input Node 2") + annotate("text", x = 0.8, y = 0.9, label = "Input Node 3") + annotate("text", x = 0.5, y = 0.6, label = "Input Layer", fontface = "bold") + annotate("rect", xmin = 0.1, xmax = 0.3, ymin = 0.4, ymax = 0.5, fill = "grey80") + annotate("rect", xmin = 0.4, xmax = 0.6, ymin = 0.4, ymax = 0.5, fill = "grey80") + annotate("rect", xmin = 0.7, xmax = 0.9, ymin = 0.4, ymax = 0.5, fill = "grey80") + annotate("text", x = 0.5, y = 0.3, label = "Output Layer", fontface = "bold") + annotate("text", x = 0.5, y = 0.1, label = "ALM + EXAM Response", fontface ="bold") print(main_plot) # Create the Gaussian curve plots for each input node input_node_1 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 0, sd = .41)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + ggtitle("Input Node 1") input_node_2 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 2, sd = .41)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + ggtitle("Input Node 2") input_node_3 <- ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = 3, sd = .41)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + ggtitle("Input Node 3") # Create the output nodes output_node_1 <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "blue") + ggtitle("Output Node 1") output_node_2 <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "blue") + ggtitle("Output Node 2") # Combine the input and output nodes using patchwork input_layer <- input_node_1 + input_node_2 + input_node_3 output_layer <- output_node_1 + output_node_2 # Create the connection matrix connection_matrix <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("segment", x = 0.2, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow()) + annotate("segment", x = 0.5, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow()) + annotate("segment", x = 0.8, xend = 0.8, y = 0.8, yend = 0.2, arrow = arrow()) # Combine the input layer, connection matrix, and output layer main_plot <- input_layer / connection_matrix / output_layer # Print the main plot print(main_plot) # Function to create an input node create_input_node <- function(mean, position) { ggplot(data.frame(x = seq(-3, 3, length.out = 100)), aes(x)) + stat_function(fun = dnorm, args = list(mean = mean, sd = .3)) + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + coord_cartesian(xlim = c(0, 3), ylim = c(0, 0.5)) + coord_flip()+ labs(title = paste("Input Node", position)) + theme(plot.title = element_text(hjust = 0.5)) } # Function to create an output node create_output_node <- function(position) { ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "grey80") + labs(title = paste("Output Node", position)) + theme(plot.title = element_text(hjust = 0.5)) } create_input_layer <- function(n) { input_plots <- map(1:n, ~create_gaussian_plot(.x)) input_layer <- wrap_plots(input_plots, ncol = 1) + coord_flip() return(input_layer) } # Create the input nodes input_nodes <- map(1:3, ~create_input_node(.x, .x)) # Create the output nodes output_nodes <- map(1:4, ~create_output_node(.x)) ## Create the connection matrix connection_matrix <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("segment", x = 1, xend = 2, y = rep(1:3, each = 4), yend = rep(1:4, times = 3), arrow = arrow()) # Combine the plots input_layer <- wrap_plots(input_nodes,byrow=FALSE,ncol=1) output_layer <- wrap_plots(output_nodes, ncol = 1) main_plot <- input_layer + connection_matrix + output_layer + plot_layout(ncol = 3) # Print the main plot print(main_plot) combined_plot <- input_layer + connection_matrix + output_layer + plot_layout(ncol = 3) print(combined_plot) print(input_layer) ``` ```{r} #| eval: false #| # Load the required libraries library(ggplot2) library(ggpp) library(patchwork) library(tidyverse) # Function to create the input layer create_input_layer <- function(n_nodes, stimulus = 0, gamma = 5) { # Create a data frame with the x values and node means df <- expand.grid(x = seq(-3, 3, length.out = 100), mean = seq(-2, 1.5, length.out = n_nodes)) %>% mutate(noisyMean=map(mean,~dnorm(.x,mean,sd=.5))) df <- df %>% mutate(distance = abs(x - stimulus), noise = rnorm(n(), sd = 0.05), max_activation = exp(-gamma * distance^2) + noise, y = max_activation * exp(-gamma * (x - mean)^2)) # Create the plot ggplot(df, aes(x, y, group = mean)) + geom_line() + geom_smooth()+ coord_cartesian(xlim = c(-2, 2), ylim = c(0, 1)) + labs(title = "Input Layer") + theme(plot.title = element_text(hjust = 0.5)) } # Function to create an output node create_output_node <- function(position) { ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("rect", xmin = -1, xmax = 1, ymin = -1, ymax = 1, fill = "grey80") + labs(title = paste("Output Node", position)) + theme(plot.title = element_text(hjust = 0.5)) } # Create the input layer input_layer <- create_input_layer(3) # Create the output nodes output_nodes <- map(1:4, ~create_output_node(.x)) # Create the connection matrix connection_matrix <- ggplot() + theme_void() + theme(plot.margin = margin(0, 0, 0, 0)) + annotate("segment", x = 1, xend = 2, y = 1:3, yend = rep(1:4, each = 3), arrow = arrow()) # Combine the plots output_layer <- wrap_plots(output_nodes, ncol = 1) main_plot <- input_layer + connection_matrix + output_layer + plot_layout(ncol = 3) # Print the main plot print(main_plot) ``` ```{r} #| eval: false # Load the visNetwork package library(visNetwork) # Define the nodes nodes <- data.frame( id = c("InputLayer", "I1", "I2", "I3", "OutputLayer", "O1", "O2", "O3", "O4", "ALM", "EXAM"), label = c("Input Layer", "Input Node 1", "Input Node 2", "Input Node 3", "Output Layer", "Output Node 1", "Output Node 2", "Output Node 3", "Output Node 4", "ALM Response", "EXAM Response"), shape = c("box", "image", "image", "image", "box", "circle", "circle", "circle", "circle", "box", "box"), image = c(NA, "gaussian_curve_input_node_1.png", "gaussian_curve_input_node_2.png", "gaussian_curve_input_node_3.png", NA, NA, NA, NA, NA, NA, NA) ) # Define the edges edges <- data.frame( from = c("InputLayer", "I1", "I1", "I1", "I1", "I2", "I2", "I2", "I2", "I3", "I3", "I3", "I3", "OutputLayer", "OutputLayer", "ALM", "EXAM"), to = c("I1", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "ALM", "EXAM", "EXAM", "ALM"), label = c("", "w1", "w2", "w3", "w4", "w5", "w6", "w7", "w8", "w9", "w10", "w11", "w12", "", "", "Pure ALM Model", "ALM with EXAM Response Component") ) # Create the network diagram visNetwork(nodes, edges, width = "100%") %>% visNodes(shapeProperties = list(useBorderWithImage = TRUE)) %>% visLayout(randomSeed = 2) # Define the nodes nodes <- data.frame( id = c("InputLayer", "I1", "I2", "I3", "OutputLayer", "O1", "O2", "O3", "O4", "ALM", "EXAM"), label = c("Input Layer", "<img src='gaussian_curve_input_node_1.png' />", "<img src='gaussian_curve_input_node_2.png' />", "<img src='gaussian_curve_input_node_3.png' />", "Output Layer", "Output Node 1", "Output Node 2", "Output Node 3", "Output Node 4", "ALM Response", "EXAM Response"), shape = c("box", "circle", "circle", "circle", "box", "circle", "circle", "circle", "circle", "box", "box") ) # Define the edges edges <- data.frame( from = c("InputLayer", "I1", "I1", "I1", "I1", "I2", "I2", "I2", "I2", "I3", "I3", "I3", "I3", "OutputLayer", "OutputLayer", "ALM", "EXAM"), to = c("I1", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "O1", "O2", "O3", "O4", "ALM", "EXAM", "EXAM", "ALM"), label = c(NA, "w1", "w2", "w3", "w4", "w5", "w6", "w7", "w8", "w9", "w10", "w11", "w12", NA, NA, "Pure ALM Model", "ALM with EXAM Response Component") ) # Create the network diagram visNetwork(nodes, edges, width = "100%") %>% visNodes(shapeProperties = list(useBorderWithImage = TRUE)) %>% visLayout(randomSeed = 2) ``` ```{r} # Load the DiagrammeR package library(DiagrammeR) # Define the graph grViz(" digraph ALM_EXAM { # Graph attributes graph [overlap = true, fontsize = 10] # Node definitions node [shape = box, fontname = Helvetica] InputLayer [label = 'Input Layer'] OutputLayer [label = 'Output Layer'] ALM [label = 'ALM Response'] EXAM [label = 'EXAM Response'] node [shape = circle, fixedsize = true, width = 0.9] I1 [label = 'Input Node 1'] I2 [label = 'Input Node 2'] I3 [label = 'Input Node 3'] O1 [label = 'Output Node 1'] O2 [label = 'Output Node 2'] O3 [label = 'Output Node 3'] O4 [label = 'Output Node 4'] # Edge definitions InputLayer -> I1 InputLayer -> I2 InputLayer -> I3 I1 -> O1 [label = 'w1'] I1 -> O2 [label = 'w2'] I1 -> O3 [label = 'w3'] I1 -> O4 [label = 'w4'] I2 -> O1 [label = 'w5'] I2 -> O2 [label = 'w6'] I2 -> O3 [label = 'w7'] I2 -> O4 [label = 'w8'] I3 -> O1 [label = 'w9'] I3 -> O2 [label = 'w10'] I3 -> O3 [label = 'w11'] I3 -> O4 [label = 'w12'] OutputLayer -> O1 OutputLayer -> O2 OutputLayer -> O3 OutputLayer -> O4 OutputLayer -> ALM OutputLayer -> EXAM ALM -> EXAM [label = 'Pure ALM Model'] EXAM -> ALM [label = 'ALM with EXAM Response Component'] } ") ``` ```{r} #| eval: false # Load the required packages library(ggplot2) library(png) # Define the Gaussian function gaussian <- function(x, mean, sd) { 1/(sd*sqrt(2*pi)) * exp(-1/2 * ((x - mean)/sd)^2) } # Define the central values for the input nodes central_values <- c(1, 2, 3) # Define the input stimulus value input_stimulus <- 2 # Generate a Gaussian curve image for each input node for (i in seq_along(central_values)) { # Define the mean and standard deviation for the Gaussian curve mean <- central_values[i] sd <- 1 # Generate the x values x <- seq(mean - 3*sd, mean + 3*sd, length.out = 100) # Generate the y values y <- gaussian(x, mean, sd) # Create the plot p <- ggplot(data.frame(x, y), aes(x, y)) + geom_line() + theme_minimal() + labs(x = "X", y = "Activation", title = paste("Input Node", i)) # Save the plot as a PNG image ggsave(paste0("gaussian_curve_input_node_", i, ".png"), plot = p, width = 4, height = 3) } ``` ```{r} #| eval: false grViz("digraph causal { # Nodes node [imagescale=true,shape = reactangle, fontname = Arial, style = filled] iv [label = 'TRT', fillcolor = '#7FC97F'] me [label = 'Mediator', shape = ellipse] dv [label = 'DLQI', fillcolor = '#7FC97F'] # Edges edge [color = black, arrowhead = normal] rankdir = LR iv -> me iv -> dv [label = 'DIRECT', fontcolor = '#7FC97F', color = '#7FC97F'] me -> dv # Graph graph [overlap = true, fontsize = 10] }") grViz("digraph causal { # Nodes node [shape = reactangle, fontname = Arial, style = filled] iv [image='gaussian_curve_input_node_1.png',label = '', fillcolor = '#7FC97F'] me [label = 'Mediator', shape = ellipse] dv [label = 'DLQI', fillcolor = '#7FC97F'] # Edges edge [color = black, arrowhead = normal] rankdir = LR iv -> me iv -> dv [label = 'DIRECT', fontcolor = '#7FC97F', color = '#7FC97F'] me -> dv # Graph graph [overlap = true, fontsize = 10] }") ``` ```{r} #| eval: false # grViz('digraph structs { # node [shape=plaintext]; # # struct1 [label=<<TABLE> # <TR><TD><IMG SRC="gaussian_curve_input_node_1.png"/></TD></TR> # <TR><TD>caption</TD></TR> # </TABLE>>]; # }') svgDataUrl = "data:image/svg+xml;base64," grViz(' digraph { node[imagescale=true, shape=circle, width=4, height=4,fontname="sans-serif",penwidth=0] a[label=<Transaction TRNA>,image="${svgDataUrl}"] b[label=<Transaction TRNB>,image="${svgDataUrl}"] c[label=<Transaction TRNC>,image="${svgDataUrl}"] a -> b a -> c }') grViz('digraph { a[image=\"gaussian_curve_input_node_1\"]; }", { images: [{ href: "gaussian_curve_input_node_1.png", width: "400px", height: "300px" }] }'); grViz(' digraph g{ node[imagescale=true, shape=circle, width=14, height=14,fontname="sans-serif",penwidth=0] I1[image="gaussian_curve_input_node_1.png", label=""]; }') grViz('digraph { ratio="fill"; size="10,10!"; margin="0,0"; node [shape=plain]; root [label=<<TABLE border="0"><TR><TD><IMG SRC="gaussian_curve_input_node_1"/></TD></TR> <TR><TD>text under</TD></TR></TABLE>>]; }') ``` ```{r} #| eval: false grViz(" digraph ALM_EXAM { # Graph attributes graph [overlap = true, fontsize = 10, rankdir = LR] # Node definitions node [shape = box, fontname = Helvetica] InputLayer [label = 'Input Layer'] OutputLayer [label = 'Output Layer'] ALM [label = 'ALM Response'] EXAM [label = 'EXAM Response'] node [shape = none, label = ''] I1 [image = 'gaussian_curve_input_node_1.png', label=''] I2 [image = 'gaussian_curve_input_node_2.png', label=''] I3 [image = 'gaussian_curve_input_node_3.png', label=''] node [shape = circle, fixedsize = true, width = 0.9] O1 [label = 'Output Node 1'] O2 [label = 'Output Node 2'] O3 [label = 'Output Node 3'] O4 [label = 'Output Node 4'] # Edge definitions InputLayer -> I1 InputLayer -> I2 InputLayer -> I3 I1 -> O1 [label = 'w1'] I1 -> O2 [label = 'w2'] I1 -> O3 [label = 'w3'] I1 -> O4 [label = 'w4'] I2 -> O1 [label = 'w5'] I2 -> O2 [label = 'w6'] I2 -> O3 [label = 'w7'] I2 -> O4 [label = 'w8'] I3 -> O1 [label = 'w9'] I3 -> O2 [label = 'w10'] I3 -> O3 [label = 'w11'] I3 -> O4 [label = 'w12'] OutputLayer -> O1 OutputLayer -> O2 OutputLayer -> O3 OutputLayer -> O4 OutputLayer -> ALM OutputLayer -> EXAM ALM -> EXAM [label = 'Pure ALM Model'] EXAM -> ALM [label = 'ALM with EXAM Response Component'] } ") ``` ```{r} DiagrammeR::grViz("digraph { graph [layout = dot, rankdir = LR] # define the global styles of the nodes. We can override these in box if we wish node [shape = rectangle, style = filled, fillcolor = Linen] data1 [label = 'Dataset 1', shape = folder, fillcolor = Beige] data2 [label = 'Dataset 2', shape = folder, fillcolor = Beige] process [label = 'Process \n Data'] statistical [label = 'Statistical \n Analysis'] results [label= 'Results'] # edge definitions with the node IDs {data1 data2} -> process -> statistical -> results }") ``` ```{mermaid} graph LR subgraph InputLayer I1["Input Node 1"] I2["Input Node 2"] I3["Input Node 3"] end subgraph OutputLayer O1["Output Node 1"] O2["Output Node 2"] O3["Output Node 3"] O4["Output Node 4"] end I1 -->|"w1"| O1 I1 -->|"w2"| O2 I1 -->|"w3"| O3 I1 -->|"w4"| O4 I2 -->|"w5"| O1 I2 -->|"w6"| O2 I2 -->|"w7"| O3 I2 -->|"w8"| O4 I3 -->|"w9"| O1 I3 -->|"w10"| O2 I3 -->|"w11"| O3 I3 -->|"w12"| O4 style InputLayer fill:#99cc99,stroke:#333,stroke-width:2px style OutputLayer fill:#cc99cc,stroke:#333,stroke-width:2px IS["Input Stimulus (X)"] --> InputLayer InputLayer -->|"Gaussian Activation (a_i)"| OutputLayer OutputLayer --> ALM["ALM Response (m(X))"] OutputLayer --> EXAM["EXAM Response (E[Y | X_i])"] ALM -->|"Pure ALM Model"| EXAM EXAM -->|"ALM with EXAM Response Component"| ALM linkStyle 0 stroke:#2ecd71,stroke-width:2px; linkStyle 1 stroke:#2ecd71,stroke-width:2px; linkStyle 2 stroke:#2ecd71,stroke-width:2px; linkStyle 3 stroke:#2ecd71,stroke-width:2px; linkStyle 4 stroke:#2ecd71,stroke-width:2px; linkStyle 5 stroke:#2ecd71,stroke-width:2px; linkStyle 6 stroke:#2ecd71,stroke-width:2px; linkStyle 7 stroke:#2ecd71,stroke-width:2px; linkStyle 8 stroke:#2ecd71,stroke-width:2px; linkStyle 9 stroke:#2ecd71,stroke-width:2px; linkStyle 10 stroke:#2ecd71,stroke-width:2px; linkStyle 11 stroke:#2ecd71,stroke-width:2px; linkStyle 12 stroke:#2ecd71,stroke-width:2px; linkStyle 13 stroke:#2ecd71,stroke-width:2px; linkStyle 14 stroke:#2ecd71,stroke-width:2px; linkStyle 15 stroke:#2ecd71,stroke-width:2px; linkStyle 16 stroke:#2ecd71,stroke-width:2px; linkStyle 17 stroke:#2ecd71,stroke-width:2px; linkStyle 0 stroke:#2ecd71,stroke-width:2px; linkStyle 1 stroke:#2ecd71,stroke-width:2px; linkStyle 2 stroke:#2ecd71,stroke-width:2px; linkStyle 3 stroke:#2ecd71,stroke-width:2px; linkStyle 4 stroke:#2ecd71,stroke-width:2px; linkStyle 5 stroke:#2ecd71,stroke-width:2px; linkStyle 6 stroke:#2ecd71,stroke-width:2px; linkStyle 7 stroke:#2ecd71,stroke-width:2px; linkStyle 8 stroke:#2ecd71,stroke-width:2px; linkStyle 9 stroke:#2ecd71,stroke-width:2px; linkStyle 10 stroke:#2ecd71,stroke-width:2px; linkStyle 11 stroke:#2ecd71,stroke-width:2px; linkStyle 12 stroke:#2ecd71,stroke-width:2px; linkStyle 13 stroke:#2ecd71,stroke-width:2px; linkStyle 14 stroke:#2ecd71,stroke-width:2px; linkStyle 15 stroke:#2ecd71,stroke-width:2px; linkStyle 16 stroke:#2ecd71,stroke-width:2px; linkStyle 17 stroke:#2ecd71,stroke-width:2px; ``` ## P5 sketching ```{ojs} P5 = require("p5") function* createSketch(sketch) { const element = DOM.element('div'); yield element; const instance = new P5(sketch, element, true); try { while (true) { yield element; } } finally { instance.remove(); } } createSketch(s => { s.setup = function() { s.createCanvas(746, 300); s.textFont('Courgette'); s.textStyle(s.BOLD); s.textAlign(s.CENTER, s.CENTER) s.button = s.createButton('clear'); s.button.mousePressed(s.clearCanvas); s.text('Click and drag to draw', s.width/2, s.height/10); }; s.draw = function() { if (s.mouseIsPressed) { s.fill(0); s.ellipse(s.mouseX, s.mouseY, 10, 10); } else { //s.fill(255); } // add text input }; // add button to clear canvas s.clearCanvas = function() { s.clear(); }; // add text // add slider } ) ``` ```{r} ```