https://chandrashaker.in/tags/multilayer-network/ Recent content in Multilayer-Network on Chandras Edu | AI, US Stocks, Courses & Jobs India Hugo en-in © 2026 Chandra Shaker · [Privacy Policy](/privacy-policy/) · [Disclaimer](/disclaimer/) · [Contact](/contact/) Mon, 01 Jun 2026 00:00:00 +0000 https://chandrashaker.in/courses/paiml/unit4/backpropagation/ Mon, 01 Jun 2026 00:00:00 +0000 https://chandrashaker.in/courses/paiml/unit4/backpropagation/ <h2 id="unit-4-navigation">Unit 4 Navigation</h2> <p><strong>← <a href="https://chandrashaker.in/courses/paiml/unit4/perceptrons/">Perceptrons</a></strong>  |  <strong>Next → <a href="https://chandrashaker.in/courses/paiml/unit4/remarks-backpropagation/">Remarks on Backpropagation</a></strong></p> <hr> <h2 id="why-multilayer-networks">Why Multilayer Networks?</h2> <p>The <a href="https://chandrashaker.in/courses/paiml/unit4/perceptrons/">perceptron</a> can only represent <strong>linear decision surfaces</strong>. Real-world problems like vowel recognition or face direction require <strong>nonlinear</strong> boundaries.</p> <p>Multilayer networks trained with Backpropagation learn <strong>rich nonlinear decision surfaces</strong> — curved, irregular boundaries that no single perceptron could produce.</p> <hr> <h2 id="451-the-sigmoid-unit">4.5.1 The Sigmoid Unit</h2> <h3 id="the-problem-with-perceptrons-in-multilayer-networks">The Problem with Perceptrons in Multilayer Networks</h3> <p>We need units that are:</p> <ol> <li><strong>Nonlinear</strong> — to express complex functions</li> <li><strong>Differentiable everywhere</strong> — to apply gradient descent</li> </ol> <p>Perceptrons fail condition 2 (hard threshold is not differentiable at the boundary). Linear units fail condition 1 (stacking linear units still gives a linear function).</p>