Nash School

What is the vanishing gradient problem in neural networks? It means the loss is not impacted by changing weights closer to input layer. If the weights do not change, the neural network stops learning and becomes bad at modelling the data.

In the figure, it is clear how the loss changes a lot when w1 is changed, and how loss remains nearly constant with w3. This is an illustration of vanishing gradients (as we move from output layer to the input layer).

🧑‍🎓📑💻

✨ Nash school London ✨

🌟Become a certified data scientist 💻💼

12-week courses 📚

Personalised curriculum 📑📙

Mentorship sessions👩🏽‍💻

International certification 🧑‍🎓

Coaches with masters and Ph.D.s ✨

Join us today! @ https://nashschool.com

Logistic Regression explained using a single pic! Check out more of our free data science content on our Instagram page: https://www.instagram.com/nashschool_datascience/

How does choice of threshold impacts true positives/false negatives in a logistic regression? We demonstrate this using pictures for two category logistic regression. 🧑‍🎓📑💻 ✨ Nash school London ✨ 🌟Become a certified data scientist 💻💼 12-week courses 📚 Personalised curriculum 📑📙 Mentorship sessions👩🏽‍💻 International certification 🧑‍🎓 Coaches with masters and Ph.D.s ✨ Join us today! @ https://nashschool.com #regression #dataanalytics #dataanalysis #statistics #datascience #datascientist #datasciencejobs #machinelearning #data #ai #optimisation #linearregression #residual #error #residuals #regularization #interaction #logistics #logisticregression #evaluation