Neil Lawrence, Maren Mahsereci, Andrei Paleyes, Mark Pullin, Cliff Facebook, Javier Gonzalez
We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.
Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.
Decision making in uncertain scenarios is an ubiquitous challenge in real world systems. Tools to deal with this challenge include simulations to gather information and statistical emulation to quantify uncertainty. The machine learning community has developed a number of methods to facilitate decision making, but so far they are scattered in multiple different toolkits, and generally rely on a fixed backend. In this paper, we present Emukit, a highly adaptable Python toolkit for enriching decision making under uncertainty. Emukit allows users to: (i) use state of the art methods including Bayesian optimization, multi-fidelity emulation, experimental design, Bayesian quadrature and sensitivity analysis; (ii) easily prototype new decision making methods for new problems. Emukit is agnostic to the underlying modeling framework and enables users to use their own custom models. We show how Emukit can be used on three exemplary case studies. 1 Intro: ML for physical processes and simulations Machine learning has demonstrated a huge success in several problems in the virtual world. Today, complex data representations can be learned using neural networks, enabling state of the art predictions in problems like image classification, natural language processing, reinforcement learning, or recommender systems. When applying machine learning in the physical world however, one faces different challenges. A key aspect of problems in robotics or climate modeling is in building models able to encapsulate physical knowledge and data. Computer simulators are widely used in scientific research and industry to analyse the behaviour of those complex systems [1]. These simulations map some knowledge about a system of interest into code that is able to reproduce real-world scenarios under varying conditions. Different types of simulators exist, ranging from computational fluid dynamics to optimization solvers. Simulators can also be used to test a system before it is built, reducing the burden and sometimes impractical limitations of real experimentation. In this paper, we define a simulator as any computer program that imitates a real-world system or process. Despite the clear advantages of using simulations, they also come with three main drawbacks: (i) a simulator can only simulate what it has been programmed to simulate, which can bias the analysis of how the system will work in the real world; (ii) a simulator can be slow and expensive to run,
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("2110.13293")
get_code_for_paper("2110.13293")
have("2110.13293")
The run record, dated, one paper per request, free:
curl https://syntology.ai/api/ran/2110.13293.json
A badge for a README (the split and the date, never a ratio):
[](https://syntology.ai/paper/2110.13293)
Connect an agent — have() is free.