{"id":245,"date":"2023-12-31T19:02:55","date_gmt":"2023-12-31T19:02:55","guid":{"rendered":"https:\/\/alandix.com\/aibook\/?page_id=245"},"modified":"2023-12-31T19:02:56","modified_gmt":"2023-12-31T19:02:56","slug":"chap09","status":"publish","type":"page","link":"https:\/\/alandix.com\/aibook\/second-edition\/toc2e\/chap09\/","title":{"rendered":"Chapter 9 \u2013 Making Sense of Machine Learning"},"content":{"rendered":"<div class=\"embedurl\" data-url=\"https:\/\/alandix.com\/books\/aibook\/content\/chaps\/chap09.html\" ><!--  Chapter 9 Making Sense of Machine Learning  -->\n\n<script>\nvar chapnos = 9;\nvar json_url = \"https:\\\/\\\/alandix.com\\\/books\\\/aibook\\\/content\\\/chaps\\\/chap09.json\";\n<\/script>\n\n\n\n\n\t<object style=\"width:100%; aspect-ratio: 10 \/ 7;\" type=\"application\/pdf\" data=\"https:\/\/alandix.com\/books\/aibook\/content\/slides-pdf\/AI-chap-09.pdf\"><\/object>\n\t<p> Download <a href=\"https:\/\/alandix.com\/books\/aibook\/content\/slides-pptx\/AI-chap-09.pptx\" download>chapter slides<\/a><\/p>\n\n\n<h3> Contents <\/h3>\n<div class=\"toc\">\n<dl>\n<dt>9.1&nbsp;&nbsp;Overview<\/dt>\n<dt>9.2&nbsp;&nbsp;The Machine Learning Process<\/dt><dd><dl>\n<dt>9.2.1&nbsp;&nbsp;Training Phase<\/dt>\n<dt>9.2.2&nbsp;&nbsp;Application Phase<\/dt>\n<dt>9.2.3&nbsp;&nbsp;Validation Phase<\/dt>\n<\/dl><\/dd>\n<dt>9.3&nbsp;&nbsp;Evaluation<\/dt><dd><dl>\n<dt>9.3.1&nbsp;&nbsp;Measures of Effectiveness<\/dt>\n<dt>9.3.2&nbsp;&nbsp;Precision--Recall Trade-off<\/dt>\n<dt>9.3.3&nbsp;&nbsp;Data for Evaluation<\/dt>\n<dt>9.3.4&nbsp;&nbsp;Multi-stage Evaluation<\/dt>\n<\/dl><\/dd>\n<dt>9.4&nbsp;&nbsp;The Fitness Landscape<\/dt><dd><dl>\n<dt>9.4.1&nbsp;&nbsp;Hill-Climbing and Gradient Descent\/Ascent<\/dt>\n<dt>9.4.2&nbsp;&nbsp;Local Maxima and Minima<\/dt>\n<dt>9.4.3&nbsp;&nbsp;Plateau and Ridge Effects<\/dt>\n<dt>9.4.4&nbsp;&nbsp;Local Structure<\/dt>\n<dt>9.4.5&nbsp;&nbsp;Approximating the Landscape<\/dt>\n<dt>9.4.6&nbsp;&nbsp;Forms of Fitness Function<\/dt>\n<\/dl><\/dd>\n<dt>9.5&nbsp;&nbsp;Dealing with Complexity<\/dt><dd><dl>\n<dt>9.5.1&nbsp;&nbsp;Degrees of Freedom and Dimension Reduction<\/dt>\n<dt>9.5.2&nbsp;&nbsp;Constraints and Dependent Features<\/dt>\n<dt>9.5.3&nbsp;&nbsp;Continuity and Learning<\/dt>\n<dt>9.5.4&nbsp;&nbsp;Multi-objective Optimisation<\/dt>\n<dt>9.5.5&nbsp;&nbsp;Partially Labelled Data<\/dt>\n<\/dl><\/dd>\n<dt>9.6&nbsp;&nbsp;Summary<\/dt>\n<\/dl><\/div>\n\n\n<h3> Glossary items referenced in this chapter <\/h3>\n<div class=\"toc\">\n<a href=\"https:\/\/alandix.com\/glossary\/aibook\/accuracy\">accuracy<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/accuracy%20measure\">accuracy measure<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/application%20phase\">application phase<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/architecture\">architecture<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/area%20under%20the%20curve\">area under the curve<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/backpropagation\">backpropagation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/base%20rate\">base rate<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/centroid\">centroid<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/clustering\">clustering<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/coherence%20of%20clusters\">coherence of clusters<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/configuration%20parameters\">configuration parameters<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/constraint%20satisfaction\">constraint satisfaction<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/constraints\">constraints<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/continuity%20in%20learning\">continuity in learning<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/cost-benefit\">cost&amp;ndash;benefit<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/cross-validation\">cross-validation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/data%20reduction\">data reduction<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/decision%20tree\">decision tree<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/deep%20neural%20network\">deep neural network<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/degrees%20of%20freedom%20%28data%29\">degrees of freedom (data)<\/a><\/strong>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/dependent%20feature\">dependent feature<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/differential%20%28calculus%29\">differential (calculus)<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/dimension%20reduction\">dimension reduction<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/ecg\">ECG <\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/energy%20landscape\">energy landscape<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/entropy\">entropy<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/explainable%20ai\">explainable AI<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/f%20score\">F score<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/false%20negative\">false negative<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/false%20positive\">false positive<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/feasible%20solution\">feasible solution<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/fitness%20function\">fitness function<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/fitness%20landscape\">fitness landscape<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/frame%20of%20video\">frame of video<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/frame%20rate\">frame rate<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/fully%20connected\">fully connected<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/generalisation\">generalisation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/genes\">genes<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/genetic%20algorithm\">genetic algorithm<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/genetic%20programming\">genetic programming<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/global%20search\">global search<\/a><\/strong>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/gradient%20ascent\">gradient ascent<\/a><\/strong>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/gradient%20descent\">gradient descent<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/hard%20threshold\">hard threshold<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/hill%20climbing%20algorithm\">hill climbing algorithm<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/hold%20out\">hold out<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/human%20labelling\">human labelling<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/id3\">ID3<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/industry%204.0\">Industry 4.0<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/intercept\">intercept<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/junk%20dna\">junk DNA<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/k-fold%20cross-validation\">k-fold cross-validation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/k-means%20algorithm\">k-means algorithm<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/kohonen%20network\">Kohonen networks<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/labelling\">labelling<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/learning%20rate\">learning rate<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/local%20maximum\">local maximum<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/local%20minima%2Fmaxima\">local minima\/maxima<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/local%20minimum\">local minimum<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/local%20search\">local search<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/locality\">locality<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/monte%20carlo%20search\">Monte Carlo search<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/multi-layer%20neural%20network\">multi-layer neural network<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/multi-objective%20optimisation\">multi-objective optimisation<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/neural%20network\">neural network<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/neuron\">neuron<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/optimal\">optimal<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/optimal%20solution\">optimal solution<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/optimisation\">optimisation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/optimisation%20criterion\">optimisation criterion<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/optimisation%20problem\">optimisation problem<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/overfitting\">overfitting<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pareto%20frontier\">Pareto frontier<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pareto%2C%20vilfredo\">Pareto, Vilfredo<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pareto-optimal\">Pareto-optimal<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/partially%20labelled%20data\">partially labelled data<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/perceptron\">perceptron<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pinch-point%20layer\">pinch-point layer<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/plateau\">plateau<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pre-processing\">pre-processing<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/precision\">precision<\/a><\/strong>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/precision-recall%20trade-off\">precision&amp;ndash;recall trade-off<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/principal%20components%20analysis\">principal components analysis<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/probability\">probability<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/pruning\">pruning<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/random%20forest\">random forest<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/recall\">recall<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/reparameterisation\">reparameterisation<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/ridge\">ridge<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/roc\">ROC<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/semi-supervised%20learning\">semi-supervised learning<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/sigmoid%20activation%20function\">sigmoid activation function<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/sigmoid%20function\">sigmoid function<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/similarity%20measure\">similarity measure<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/simulated%20annealing\">simulated annealing<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/smoothing\">smoothing<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/soft%20constraints\">soft constraints<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/speech%20recognition\">speech recognition<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/statistical%20techniques\">statistical techniques<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/supervised%20learning\">supervised learning<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/threshold\">threshold<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/training%20phase\">training phase<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/true%20negative\">true negative<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/true%20positive\">true positive<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/unsupervised%20learning\">unsupervised learning<\/a>, <strong><a href=\"https:\/\/alandix.com\/glossary\/aibook\/validation%20phase\">validation phase<\/a><\/strong>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/version-space%20algorithm\">version-space algorithm<\/a>, <a href=\"https:\/\/alandix.com\/glossary\/aibook\/visualisation\">visualisation<\/a><\/div>\n\n\n\n\n<\/div>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":2,"featured_media":0,"parent":221,"menu_order":9,"comment_status":"closed","ping_status":"closed","template":"","meta":{"_themeisle_gutenberg_block_has_review":false,"footnotes":""},"class_list":["post-245","page","type-page","status-publish","hentry"],"_links":{"self":[{"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/pages\/245","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/comments?post=245"}],"version-history":[{"count":3,"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/pages\/245\/revisions"}],"predecessor-version":[{"id":299,"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/pages\/245\/revisions\/299"}],"up":[{"embeddable":true,"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/pages\/221"}],"wp:attachment":[{"href":"https:\/\/alandix.com\/aibook\/wp-json\/wp\/v2\/media?parent=245"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}