GANs: The Case of Predictive Business Process
summary
Uses a GAN with LSTM generator/discriminator to predict the next user action on dell.com, trained alternately with binary cross-entropy loss; evaluation uses weighted accuracy and F1 for sequences (FID only discussed for images). A proof of concept on hundreds of thousands of records produces a significant improvement over baseline LSTM according to the presenter, yields a working POC but remains POC not production.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:08GAN applications
- 4:43Generative vs discriminative
- 6:27Discriminator and generator
- 9:56Min-max loss and training
- 12:47Training process and code
- 18:05Evaluating with FID
- 22:01Dell use case
- 24:29GANs with LSTMs
- 25:36Results and impact
- 26:38Q&A
related talks
53mGenerative AI: Overview and use cases
15mHarnessing the Power of Generative AI for Intelligent Applications
06mGAN-based Style Transformation to Improve Gesture-recognition Accuracy
14mGenerative AI models: Practical Business Applications
22mGenerative AI & its Transformation of Business, Research, and Society
49mGenerative AI: Concepts, Tools, and Applications for Data Professionals
01h 00mTutorial on Deep Learning IV
48mUnveiling the secrets of a machine learning model