update overview and add torch operators (#161)

mlr-org · Aug 16, 2024 · 9a08132 · 9a08132
1 parent 896d049
commit 9a08132
Show file tree

Hide file tree

Showing 7 changed files with 71 additions and 18 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -41,6 +41,7 @@ Imports:
     mlr3spatiotempcv,
     mlr3tuning,
     mlr3tuningspaces,
+    mlr3torch,
     mlr3verse,
     mlr3viz,
     paradox,
@@ -49,10 +50,11 @@ Imports:
     reactable,
     rmarkdown,
     rprojroot,
-    stringi
+    stringi,
+    torchvision
 Remotes:
     mlr-org/mlr3extralearners,
-    mlr-org/mlr3fda,
+    mlr-org/mlr3torch,
     mlr-org/mlr3proba
 Encoding: UTF-8
 Roxygen: list(markdown = TRUE)

diff --git a/R/zzz.R b/R/zzz.R
@@ -8,7 +8,7 @@
 
 db = new.env()
 db$index = c("base", "utils", "datasets", "data.table", "stats", "batchtools")
-db$hosted = c("paradox", "mlr3misc", "mlr3", "mlr3data", "mlr3db", "mlr3proba", "mlr3pipelines", "mlr3learners", "mlr3filters", "bbotk", "mlr3tuning", "mlr3viz", "mlr3fselect", "mlr3cluster", "mlr3spatiotempcv", "mlr3spatial", "mlr3extralearners", "mlr3tuningspaces", "mlr3hyperband", "mlr3mbo", "mlr3verse", "mlr3benchmark", "mlr3oml", "mlr3batchmark", "mlr3fairness", "mlr3fda")
+db$hosted = c("paradox", "mlr3misc", "mlr3", "mlr3data", "mlr3db", "mlr3proba", "mlr3pipelines", "mlr3learners", "mlr3filters", "bbotk", "mlr3tuning", "mlr3viz", "mlr3fselect", "mlr3cluster", "mlr3spatiotempcv", "mlr3spatial", "mlr3extralearners", "mlr3tuningspaces", "mlr3hyperband", "mlr3mbo", "mlr3verse", "mlr3benchmark", "mlr3oml", "mlr3batchmark", "mlr3fairness", "mlr3fda", "mlr3torch")
 
 lgr = NULL
 

diff --git a/mlr-org/_quarto.yml b/mlr-org/_quarto.yml
@@ -40,6 +40,7 @@ website:
            - text: "---"
            - pipeops.qmd
            - graphs.qmd
+           - torch_pipeops.qmd
            - text: "---"
            - tuners.qmd
            - tuning_spaces.qmd

diff --git a/mlr-org/ecosystem.qmd b/mlr-org/ecosystem.qmd
@@ -72,21 +72,23 @@ content = rowwise_table(
     "data",            "mlr3data",          "stable",       "https://mlr3data.mlr-org.com/",              "Data sets and tasks",
     "learners",        "mlr3learners",      "stable",       "https://mlr3learners.mlr-org.com/",          "Essential learners for mlr3, maintained by the mlr-org team",
     "learners",        "mlr3extralearners", "stable",       "https://mlr3extralearners.mlr-org.com/",     "Extra learners for mlr3, implemented by the community",
-    "learners",        "mlr3torch",         "experimental", "https://mlr3torch.mlr-org.com",              "Deep learning with Torch",
+    "learners",        "mlr3torch",         "stable",       "https://mlr3torch.mlr-org.com",              "Deep learning with torch",
     "analysis",        "mlr3viz",           "stable",       "https://mlr3viz.mlr-org.com/",               "Visualizations for tasks, predictions, resample results and benchmarks",
     "analysis",        "mlr3benchmark",     "stable",       "https://mlr3benchmark.mlr-org.com/",         "Analysis and tools for benchmarking",
+    "analysis",        "mlr3inferr",        "stable",       "https://mlr3inferr.mlr-org.com/",            "Statistical methods for inference on the generalization error",
+    "analysis",        "mlr3summary",       "stable",       "https://github.com/mlr-org/mlr3summary",     "Summary methods for mlr3",
     "tasks",           "mlr3spatiotempcv",  "stable",       "https://mlr3spatiotempcv.mlr-org.com/",      "Spatiotemporal resampling and visualization methods",
     "tasks",           "mlr3cluster",       "stable",       "https://mlr3cluster.mlr-org.com/",           "Cluster analysis",
     "tasks",           "mlr3proba",         "stable",       "https://mlr3proba.mlr-org.com/",             "Probabilistic predictions",
     "tasks",           "mlr3temporal",      "experimental", "https://mlr3temporal.mlr-org.com/",          "Time series analysis",
     "tasks",           "mlr3spatial",       "stable",       "https://mlr3spatial.mlr-org.com/",           "Spatial data backends and prediction functions",
-    "tasks",           "mlr3fda",           "experimental", "https://mlr3fda.mlr-org.com",                "Functional Data Analysis",
+    "tasks",           "mlr3fda",           "stable",       "https://mlr3fda.mlr-org.com",                "Functional Data Analysis",
     "tasks",           "mlr3fairness",      "stable",       "https://mlr3fairness.mlr-org.com/",          "Fairness in Machine Learning",
-    "utils",           "paradox",           "stable",       "https://paradox.mlr-org.com/",               "Universal parameter space description and tools",
-    "utils",           "mlr3misc",          "stable",       "https://mlr3misc.mlr-org.com/",              "Miscellaneous helper functions for mlr3",
-    "utils",           "mlr3measures",      "stable",       "https://mlr3measures.mlr-org.com/",          "Performance measures for supervised learning",
-    "parallelization", "mlr3batchmark",     "experimental", "https://github.com/mlr-org/mlr3batchmark/",  "Connector between mlr3 and batchtools",
-    "parallelization", "rush",              "experimental", "https://github.com/mlr-org/rush",            "Asynchronous parallelization"
+    "other",           "paradox",           "stable",       "https://paradox.mlr-org.com/",               "Universal parameter space description and tools",
+    "other",           "mlr3misc",          "stable",       "https://mlr3misc.mlr-org.com/",              "Miscellaneous helper functions for mlr3",
+    "other",           "mlr3measures",      "stable",       "https://mlr3measures.mlr-org.com/",          "Performance measures for supervised learning",
+    "parallelization", "mlr3batchmark",     "stable",       "https://github.com/mlr-org/mlr3batchmark/",  "Connector between mlr3 and batchtools",
+    "parallelization", "rush",              "stable",       "https://github.com/mlr-org/rush",            "Asynchronous parallelization"
 )
 # add CRAN link
 content[, on_cran := name %in% rownames(available.packages(repos = "https://cloud.r-project.org"))]
@@ -105,15 +107,15 @@ setkeyv(content, c("category", "status"))
 #| echo: false
 package_grid(content[.("core")])
 ```
-
+ 
 ## <i class="bi-pen"></i> Learners
 
 ```{r}
 #| echo: false
 package_grid(content[.("learners")])
 ```
 
-## <i class="bi-sliders"></i> Optimization
+## <i class="bi-sliders"></i> Tuning & Optimization
 
 ```{r}
 #| echo: false
@@ -149,11 +151,11 @@ package_grid(content[.("data")])
 package_grid(content[.("analysis")])
 ```
 
-## <i class="bi-balloon"></i> Utilities
+## <i class="bi-balloon"></i> Other
 
 ```{r}
 #| echo: false
-package_grid(content[.("utils")])
+package_grid(content[.("other")])
 ```
 
 ## <i class="bi-share"></i> Parallelization

diff --git a/mlr-org/learners.qmd b/mlr-org/learners.qmd
@@ -15,26 +15,29 @@ library("mlr3learners")
 library("mlr3extralearners")
 library("mlr3proba")
 library("mlr3cluster")
+library("mlr3torch")
 ```
 
 To keep the dependencies on other packages reasonable, the base package mlr3 only ships with with regression and classification trees from the `r ref_pkg("rpart")` package and some learners for debugging.
 A subjective selection of implementations for essential ML algorithms can be found in `r ref_pkg("mlr3learners")` package.
 Survival learners are provided by `r ref_pkg("mlr3proba")`, cluster learners via `r ref_pkg("mlr3cluster")`.
 Additional learners, including some learners which are not yet to be considered stable or which are not available on CRAN, are connected via the `r ref_pkg("mlr3extralearners")` package.
-
+For neural networks, see the `r ref_pkg("mlr3torch")` extension.
 
 ```{r}
 #| echo: false
+#| warning: false
 #| column: page
 
 content = as.data.table(mlr_learners, objects = TRUE)
 content[, base_package := map(object, function(x) strsplit(x$man, "::", TRUE)[[1]][1])]
 content[, packages := pmap(list(packages, base_package), function(x, y) setdiff(x, c(y, "mlr3")))]
+learners = set_names(content$object, content$key)
 content[, `:=`(object = NULL, task_type = NULL)]
 
 ref_cell = function(value) {
   if (!length(value)) return(NULL)
-  mlr3website::ref(sprintf("%s_%s", "mlr_learners", value), text = value, format = "htmltools")
+  mlr3website::ref(learners[[value]]$man, text = value, format = "htmltools")
 }
 
 reactable(content, columns = list(

diff --git a/mlr-org/pipeops.qmd b/mlr-org/pipeops.qmd
@@ -13,6 +13,7 @@ toc: false
 library("mlr3verse")
 library("mlr3proba")
 library("mlr3fda")
+library("mlr3torch")
 ```
 
 Objects of class `r ref("PipeOp")` are the building blocks to compose linear machine learning pipelines and non-linear `r ref("Graph", text = "Graphs")`.
@@ -23,10 +24,10 @@ The base objects are implemented in `r ref_pkg("mlr3pipelines")`.
 #| column: page
 
 mlr_pipeops$remove("breslowcompose") # workaround
-content = as.data.table(mlr_pipeops, objects = TRUE)[, .(key, label, packages, tags)]
+content = as.data.table(mlr_pipeops)
 content = content[key %nin% c("crankcompose", "distrcompose")] # deprecated, no help page
 content[, packages := map(packages, function(value) setdiff(value, "mlr3pipelines"))]
-
+content = content[, .(key, label, packages, tags)]
 
 ref_cell = function(value) {
   if (!length(value)) return(NULL)

diff --git a/mlr-org/torch_pipeops.qmd b/mlr-org/torch_pipeops.qmd
@@ -0,0 +1,44 @@
+---
+sidebar: false
+toc: false
+---
+
+# Torch Pipeline Operators
+
+{{< include _setup.qmd >}}
+
+```{r}
+#| include: false
+
+library("mlr3pipelines")
+library("mlr3torch")
+```
+
+`r ref_pkg("mlr3torch")` supports a number of `r ref("PipeOp")` building blocks to define neural networks, including preprocessing and data augmentation steps.
+
+```{r}
+#| echo: false
+#| column: page
+
+content = as.data.table(mlr_pipeops)
+content = content[key %nin% c("crankcompose", "distrcompose")] # deprecated, no help page
+content[, packages := map(packages, function(value) setdiff(value, "mlr3pipelines"))]
+content = content[, .(key, label, packages, tags)]
+content = content[mlr3misc::map_lgl(packages, function(ps) "torch" %in% ps), ]
+
+ref_cell = function(value) {
+  if (!length(value)) return(NULL)
+  mlr3website::ref(sprintf("%s_%s", "mlr_pipeops", value), text = value, format = "htmltools")
+}
+
+reactable(content, columns = list(
+    key = colDef(name = "Key", cell = ref_cell),
+    label = colDef(name = "Label"),
+    packages = colDef(name = "Packages", cell = package_list_cell),
+    tags = colDef(name = "Tags", cell = list_cell)
+  ),
+  searchable = TRUE,
+  pagination = FALSE,
+  highlight = TRUE
+)
+```