Update learning back button

learning/courses/basics-of-quantum-information/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Basics of quantum information",
   "collapsed": true,
   "children": [

learning/courses/foundations-of-quantum-error-correction/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Foundations of quantum error correction",
   "collapsed": true,
   "children": [

learning/courses/fundamentals-of-quantum-algorithms/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Fundamentals of quantum algorithms",
   "collapsed": true,
   "children": [

learning/courses/general-formulation-of-quantum-information/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "General formulation of quantum information",
   "collapsed": true,
   "children": [

learning/courses/quantum-business-foundations/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Quantum business foundations",
   "collapsed": true,
   "children": [

learning/courses/quantum-chem-with-vqe/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Quantum chemistry with VQE",
   "collapsed": true,
   "children": [

learning/courses/quantum-computing-in-practice/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Quantum computing in practice",
   "collapsed": true,
   "children": [

learning/courses/quantum-diagonalization-algorithms/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Quantum diagonalization algorithms",
   "collapsed": true,
   "children": [

learning/courses/quantum-machine-learning/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Quantum machine learning",
   "collapsed": true,
   "children": [

learning/courses/quantum-machine-learning/classical-ml-review.ipynb

@@ -20,15 +20,15 @@
    "source": [
     "## Types of machine learning\n",
     "\n",
-    "### Supervision\n",
-    "\n",
     "As a simple definition, machine learning is a collection of algorithms that analyze and draw inferences from patterns and relationships in data. Broadly speaking, machine learning algorithms can be grouped into three main categories depending on the type of data involved and how algorithms learn without being explicitly programmed:\n",
     "1. __Supervised learning:__ In supervised learning, the data that is used to train the model is labeled. The goal of these algorithms is to learn the relationship between data and their corresponding labels or outputs and to generalize this to unseen data. Common tasks in this class are classification and regression.\n",
     "2. __Unsupervised learning:__ In contrast to supervised learning, unsupervised learning uses unlabeled data to train the machine learning model. The goal of such algorithms is to discover hidden patterns and structure in data. Some algorithms in this class are clustering and dimensionality reduction algorithms. Some generative models such as generative adversarial networks and variational autoencoders can also be considered in this category.\n",
     "3. __Reinforcement learning:__ Algorithms in this machine learning category are defined by an agent which interacts with an environment. The agent takes actions and receives feedback from its environment in the form of rewards and punishments. Eventually through this feedback mechanism, the agent learns to take the correct set of actions to perform a specific task.\n",
     "\n",
     "![A diagrammatic representation of supervised and unsupervised learning.](/learning/images/courses/quantum-machine-learning/classical-ml-review/qml-cr-background-sup-unsup.avif)\n",
     "\n",
+    "The left image shows two categories of labeled data as in supervised learning. In this case, the categories are linearly separable. The right image shows clusters of data. In an unsupervised learning task, these data would not initially be labeled and the algorithm would study the distribution, perhaps looking for clusters. For the purposes of visualizing example clusters the algorithm might identify, the data points have now been labeled. A key difference between the two is that the supervised learning process starts with the data already labeled and the unsupervised process starts with unlabeled data, even if the data are labeled at the end.\n",
+    "\n",
     "\n",
     "### Introducing “quantum” to machine learning\n",
     "\n",

learning/courses/quantum-machine-learning/introduction.ipynb

@@ -44,6 +44,8 @@
     "\n",
     "![QML_CR_background_Sup_Unsup.avif](/learning/images/courses/quantum-machine-learning/introduction/qml-cr-background-sup-unsup.avif)\n",
     "\n",
+    "The left image shows two categories of labeled data as in supervised learning. In this case, the categories are linearly separable. The right image shows clusters of data. In an unsupervised learning task, these data would not initially be labeled and the algorithm would study the distribution, perhaps looking for clusters. For the purposes of visualizing example clusters the algorithm might identify, the data points have now been labeled. A key difference between the two is that the supervised learning process starts with the data already labeled and the unsupervised process starts with unlabeled data, even if the data are labeled at the end.\n",
+    "\n",
     "Those with background in machine learning will already know that many solution methods involve mapping data into higher-dimensional spaces. This is especially well-explored in the context of kernels. As a brief reminder, sometimes data may be separable into categories by a line, plane, or hyperplane (we will often simply say \"hyperplane\" for compactness), in the same number of dimensions as the data are given. This is shown in the first image above. Other times, data may not be separable by a hyperplane in those dimensions, as shown in the second image. But there can still be structure to the data that can be exploited in a mapping to higher dimensions, which then leaves the data separable in that higher-dimensional space. This is illustrated in the mapping of the 2D data with circular symmetry into the 3D space in which the data points are arranged along a paraboloid surface.\n",
     "\n",
     "![QML_CR_background_2D-3D.avif](/learning/images/courses/quantum-machine-learning/introduction/qml-cr-background-2d-3d.avif)\n",

learning/courses/quantum-safe-cryptography/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "A practical introduction to quantum-safe cryptography",
   "collapsed": true,
   "children": [

learning/courses/utility-scale-quantum-computing/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Utility-scale quantum computing",
   "collapsed": true,
   "children": [

learning/courses/variational-algorithm-design/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/courses",
   "title": "Variational algorithm design",
   "collapsed": true,
   "children": [

learning/modules/computer-science/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/modules",
   "title": "Qiskit in the classroom - computer science",
   "collapsed": true,
   "children": [

learning/modules/quantum-mechanics/_toc.json

@@ -1,5 +1,5 @@
 {
-  "parentUrl": "/learning",
+  "parentUrl": "/learning/modules",
   "title": "Qiskit in the classroom - quantum mechanics",
   "collapsed": true,
   "children": [