Slides 7 & 8

aronwalsh · Feb 12, 2024 · 68ec0b6 · 68ec0b6
1 parent f32f05c
commit 68ec0b6
Show file tree

Hide file tree

Showing 3 changed files with 3 additions and 3 deletions.
diff --git a/Lecture6.ipynb b/Lecture6.ipynb
@@ -201,7 +201,7 @@
     "np.random() generates pseudorandom numbers. The algorithm starts with an initial seed value and then produces a sequence of numbers that appears random. However, since the process is deterministic, providing the same seed will result in the same sequence of \"random\" numbers. 42 is simply a science fiction reference. \n",
     "</details>\n",
     "\n",
-    "That was fun, but a lone perceptron is limited in terms of the behaviour it can represent. How could we simulate a more realistic phase diagram that contains more than two regions? Let's code a neural network!\n",
+    "That was fun, but a lone perceptron is limited in terms of the behaviour it can represent. How could we simulate more realistic phase behaviour? Let's code a neural network!\n",
     "\n",
     "Here the sigmoid activation function is used, which can be represented mathematically as:\n",
     "\n",
@@ -323,7 +323,7 @@
     "tags": []
    },
    "source": [
-    "## Image processing\n",
+    "## Image processing\n",
     "\n",
     "Many characterisation techniques in materials science produce image data. Researchers can save time and perform advanced experiments (e.g. active learning) and analysis (e.g. feature identification) using codes. While \"point and click\" software solutions exist for image processing, they are often expensive and difficult to customise and automate for large datasets.\n",
     "\n",
@@ -1187,7 +1187,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.13"
+   "version": "3.9.18"
   }
  },
  "nbformat": 4,

diff --git a/slides/MLforMaterials_Lecture7_Build.pdf b/slides/MLforMaterials_Lecture7_Build.pdf
diff --git a/slides/MLforMaterials_Lecture8_AI.pdf b/slides/MLforMaterials_Lecture8_AI.pdf