Task-based End-to-end Model Learning
By Priya Lekha Donti, Brandon Amos, J. Kolter. 2017
25 citations recorded by the source corpus.
As machine learning techniques have become more ubiquitous, it has become common to see machine learning prediction algorithms operating within some larger process. However, the criteria by which we train machine learning algorithms often differ from the ultimate criteria on which we evaluate them. This paper proposes an end-to-end approach for learning probabilistic machine learning models within the context of stochastic programming, in a manner that directly captures the ultimate task-based objective for which they will be used. We then present two experimental evaluations of the proposed approach, one as applied to a generic inventory stock problem and the second to a real-world electrical grid scheduling task. In both cases, we show that the proposed approach can outperform both a traditional modeling approach and a purely black-box policy optimization approach.
Subfields: Artificial Intelligence, Electrical and Electronic Engineering, Management Science and Operations Research, Optimization & Training
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Electrical and Electronic Engineering
In Management Science and Operations Research
In Optimization & Training
Citations in the corpus
- Deep Reinforcement Learning (2018, unknown): “Donti et al. (2017) propose differentiating through a stochastic programming optimisation using a probabilistic model to learn model parameters with respect to their true objective rather than a maximum likelihood surrogate.”
- Smart "Predict, then Optimize" (2017, methodology): “Donti et al. (2017) proposes a heuristic to address a more general setting than that of Kao et al. (2009), and also focus on the case of quadratic optimization. These works also bypass issues of non-uniqueness of solutions of the nominal problem (since their problem is strongly convex), which must b”