Differentiable programming is the latest descendant of deep learning. It has unlocked a series of challenges which were previously seen as “unsolvable” and has paved the way for considerable progress and superior numerical results in the world of Supply Chains. In this episode of LokadTV we tackle this extremely exciting new development and learn how this major breakthrough means we can now perform optimizations, even when faced with high levels of uncertainty.
In supply chains there are a whole series of unsolved problems that the industry has been struggling with for years. We learn how solutions to the big AI problems, such as voice recognition, image analysis and autonomous vehicles, can be applied to specific supply chain products and why the present time has been the better moment to revisit these failures.
We explore how differentiable programming provides a joint resolution to these problems and makes it possible to mix factors that are logically so different. Finally, we discuss what this breakthrough in large-scale numerical optimisation means for future supply chains and how considerable progress and superior numerical results can be made.
But there are still many “wicked” problems that remain unsolvable, despite these technological leaps and bounds. Problems that would “take Skynet to solve”. So how far away are we from a Terminator-esque future with robots taking over the world? Watch and find out!
00:28 Today, we are going to talk about tackling some unsolved problems. What kind of problems are we talking about?
01:48 What sort of problems has the Supply Chain industry been struggling with for years?
04:20 Can we combine existing methods to get something satisfactory?
09:22 Where does it work really well in the real world? What are the classic challenges that it is really good at solving?
13:18 Which are the problems in the Supply Chain world which Differentiable Programming cannot solve?
17:01 Why was now a better time to revisit these unsolvable problems?
20:07 Is that growth going to continue or not?
21:40 What are the benefits of Differentiable Programming for Supply Chain practitioners? How is it going to change the way in which they are approaching problems?