English suggestions for autoencoder

Author: AnnePicus

algorithms/qml/quantum_autoencoder/quantum_autoencoder.ipynb

@@ -13,11 +13,13 @@
    "id": "201ce387",
    "metadata": {},
    "source": [
-    "### Classical encoders:\n",
+    "## Encoder Types\n",
+    "\n",
+    "### Classical Encoders\n",
     "Refer to encoding/compressing classical data into a smaller sized data via deterministic algorithm. For example, JPEG is essentially an algorithm which compresses images into a smaller sized images.\n",
     "\n",
-    "### Classical auto-encoders:\n",
-    "One can use machine-learning technics and train a variational network for compressing data. In general, an auto-encoder network looks as follows:\n",
+    "### Classical Autoencoders\n",
+    "One can use machine-learning technics and train a variational network for compressing data. In general, an autoencoder network looks as follows:\n",
     "\n",
     "<center>\n",
     "<img src=\"https://docs.classiq.io/resources/Autoencoder_structure.png\" style=\"width:50%\">\n",
@@ -35,8 +37,8 @@
    "id": "5be9dbfd",
    "metadata": {},
    "source": [
-    "### Quantum auto-encoders:\n",
-    "In a similar fashion to the classical counterpart, quantum auto-encoder refers to \"compressing\" quantum data stored initially on $n$ qubits into a smaller quantum register of $m<n$ qubits, via variational circuit. However, quantum computing is reversibale, and thus qubits cannot be \"erased\". Therefore, alternatively, a quantum autoencoder tries to acheive the following transformation from uncoded quantum register of size $n$ to a coded one of size $m$:\n",
+    "### Quantum Autoencoders\n",
+    "In a similar fashion to the classical counterpart, quantum autoencoder refers to \"compressing\" quantum data stored initially on $n$ qubits into a smaller quantum register of $m<n$ qubits, via variational circuit. However, quantum computing is reversibale, and thus qubits cannot be \"erased\". Therefore, alternatively, a quantum autoencoder tries to acheive the following transformation from uncoded quantum register of size $n$ to a coded one of size $m$:\n",
     "$$\n",
     "|\\psi\\rangle_n \\rightarrow |\\psi'\\rangle_m|0\\rangle_{n-m}\n",
     "$$\n",
@@ -53,11 +55,11 @@
    "id": "e40c41f0",
    "metadata": {},
    "source": [
-    "# Training of quantum auto encoders\n",
+    "## Training of Quantum Autoencoders\n",
     "\n",
     "To train a quantum auto encoder one should define a proper cost function. Below we propose two common approaches, one using a swap test and the other uses Hamiltonian measurements. We focus on the swap test case, and comment on the other approach at the end of this notebook.\n",
     "\n",
-    "## The swap test\n",
+    "### The Swap Test\n",
     "\n",
     "The swap test is a quantum function which checks the overlap between two quantum states: the inputs of the function are two quantum registers of the same size, $|\\psi_1\\rangle, \\,|\\psi_2\\rangle$, and it returns as an output a single \"test\" qubit whose state encodes the overlap between the two inputs: $|q\\rangle_{\\rm test} = \\alpha|0\\rangle + \\sqrt{1-\\alpha^2}|1\\rangle$, with\n",
     "$$\n",
@@ -81,7 +83,7 @@
    "id": "443577d1",
    "metadata": {},
    "source": [
-    "## Quantum neural network for quantum auto encoder\n",
+    "### Quantum Neural Networks for Quantum Autoencoders\n",
     "\n",
     "The quantum auto encoder can be built as a quantum neural network, having the following three parts:\n",
     "\n",
@@ -102,7 +104,7 @@
    "id": "6a685c3b",
    "metadata": {},
    "source": [
-    "# Pre-user-defined functions which will be used to construct the quantum layer\n",
+    "## Pre-user-defined Functions That Construct the Quantum Layer\n",
     "\n",
     "As a first step we build some user-defined functions which allow us flexible modeling. We have three functions:\n",
     "1. `angle_encoding`: This function loads data of size `num_qubits` on `num_qubits` qubits via RY gates. It has an output port named `qpv`.\n",
@@ -208,7 +210,7 @@
    "id": "493d4499",
    "metadata": {},
    "source": [
-    "# An example: auto encoder for domain wall data"
+    "## Example: Autoencoder for Domain Wall Data"
    ]
   },
   {
@@ -224,7 +226,7 @@
    "id": "535d9181",
    "metadata": {},
    "source": [
-    "## The data"
+    "### The Data"
    ]
   },
   {
@@ -263,7 +265,7 @@
    "id": "0a09e977",
    "metadata": {},
    "source": [
-    "## The quantum program"
+    "### The Quantum Program"
    ]
   },
   {
@@ -382,7 +384,7 @@
    "id": "5a41ccaa",
    "metadata": {},
    "source": [
-    "## The network\n",
+    "### The Network\n",
     "\n",
     "The network for training contains only a quantum layer. The corresponding quantum program was already defined above, what is left is to define some execution preferences and the classical post-process. The classical output is defined as $1-\\alpha^2$, with $\\alpha$ being the probability of the test qubit to be at state 0."
    ]
@@ -494,7 +496,7 @@
    "id": "7590c55d",
    "metadata": {},
    "source": [
-    "## Creating dataset\n",
+    "### Creating the Dataset\n",
     "\n",
     "The cost function we would like to minimize is $|1-\\alpha^2|$ for all our training data. Looking at the qlayer output this means that we should define the corresponding labels as $0$."
    ]
@@ -551,7 +553,7 @@
    "id": "608173a0",
    "metadata": {},
    "source": [
-    "## Define the training"
+    "### Defining the Training"
    ]
   },
   {
@@ -596,7 +598,7 @@
    "id": "ee98061f",
    "metadata": {},
    "source": [
-    "## Setting some hyper-parameters\n",
+    "### Setting Hyper-parameters\n",
     "\n",
     "The L1 loss function fits the intended cost function we aim to minimize."
    ]
@@ -625,7 +627,7 @@
    "id": "e38d76f1",
    "metadata": {},
    "source": [
-    "## Training\n",
+    "### Training\n",
     "\n",
     "In this demo we will initialize the network with trained parameters, and run only 1 epoch for demonstration. A reasonable training with the above hyper-parameters can be achieved with $\\sim 40$ epochs. To train the network from the beginning uncomment the following code line:"
    ]
@@ -685,7 +687,7 @@
    "id": "7fb11e18",
    "metadata": {},
    "source": [
-    "## Verification\n",
+    "### Verification\n",
     "\n",
     "Once we trained our network, we can build a new network with the trained variables. We can thus verify our encoder by taking only the encoding block, changing post_process, etc.\n",
     "\n",
@@ -708,7 +710,7 @@
    "id": "812d2bc5",
    "metadata": {},
    "source": [
-    "### We start with building the quantum layer for the validator"
+    "### Building the Quantum Layer for the Validator"
    ]
   },
   {
@@ -840,7 +842,9 @@
    "id": "7f3f9983",
    "metadata": {},
    "source": [
-    "### Next, we define the classical output of the network. For the validator post-process we take the output with the maximal counts."
+    "### Defining the Classical Output of the Network \n",
+    "\n",
+    "For the validator postprocessing, we take the output with the maximal counts:"
    ]
   },
   {
@@ -890,7 +894,7 @@
    "id": "39a94e36",
    "metadata": {},
    "source": [
-    "### We create the network and assign the trained parameters"
+    "### Creating the Network and Assigning the Trained Parameters"
    ]
   },
   {
@@ -989,7 +993,7 @@
    "id": "6dca60d3",
    "metadata": {},
    "source": [
-    "# Usage for anomaly detection"
+    "## Detecting Anomalies"
    ]
   },
   {