Equation 4-18. Logistic cost function partial derivatives do = ] (467.39) 9) Show that Equation 4-18 (Geron: Hands-on Machine Learning) is correct by finding out all partial derivatives in the computation graphs for the two y values after a single data point (x,y) is seen. KO) = -log("") + (1 - yog(1 $9)] Equation 4-13. Logistic Regression model estimated probability P = g(x) = m(p - x) Computation Graph for (x, y=1) d430 dija alap mul Ot log OL/op Op/oz OL/Oz oz/de ol/de dz/00 dp/dz Remember that (logu)' = (1/uju' and (o[u]] = 0[u) (1 - {u}}u. L=J() = -log o when y = 1. Use p for ^p. Computation Graph for (x, y=0) OL/de al/a du op d/a mul O. OL/og og/op OL/op Op/oz OL/Oz Oz/de OL/0E dz/de dp/d2 dolop Here we have q = 1-p and L= J(e) = -log[1 - p) when y=0. Equation 4-18. Logistic cost function partial derivatives do = ] (467.39) 9) Show that Equation 4-18 (Geron: Hands-on Machine Learning) is correct by finding out all partial derivatives in the computation graphs for the two y values after a single data point (x,y) is seen. KO) = -log("") + (1 - yog(1 $9)] Equation 4-13. Logistic Regression model estimated probability P = g(x) = m(p - x) Computation Graph for (x, y=1) d430 dija alap mul Ot log OL/op Op/oz OL/Oz oz/de ol/de dz/00 dp/dz Remember that (logu)' = (1/uju' and (o[u]] = 0[u) (1 - {u}}u. L=J() = -log o when y = 1. Use p for ^p. Computation Graph for (x, y=0) OL/de al/a du op d/a mul O. OL/og og/op OL/op Op/oz OL/Oz Oz/de OL/0E dz/de dp/d2 dolop Here we have q = 1-p and L= J(e) = -log[1 - p) when y=0