ML p(r)ior | Better safe than sorry: Risky function exploitation through safe optimization

Better safe than sorry: Risky function exploitation through safe optimization

2016-02-02
Exploration-exploitation of functions, that is learning and optimizing a mapping between inputs and expected outputs, is ubiquitous to many real world situations. These situations sometimes require us to avoid certain outcomes at all cost, for example because they are poisonous, harmful, or otherwise dangerous. We test participants' behavior in scenarios in which they have to find the optimum of a function while at the same time avoid outputs below a certain threshold. In two experiments, we find that Safe-Optimization, a Gaussian Process-based exploration-exploitation algorithm, describes participants' behavior well and that participants seem to care firstly whether a point is safe and then try to pick the optimal point from all such safe points. This means that their trade-off between exploration and exploitation can be seen as an intelligent, approximate, and homeostasis-driven strategy.
PDF

Highlights - Most important sentences from the article

Login to like/save this paper, take notes and configure your recommendations

Related Articles

2018-07-04

A neural network (NN) is a parameterised function that can be tuned via gradient descent to approxim… show more
PDF

Highlights - Most important sentences from the article

2016-11-17

In recent years deep reinforcement learning (RL) systems have attained superhuman performance in a n… show more
PDF

Highlights - Most important sentences from the article

2018-07-08
1807.02811 | stat.ML

Bayesian optimization is an approach to optimizing objective functions that take a long time (minute… show more
PDF

Highlights - Most important sentences from the article

2016-02-14

Robotic algorithms typically depend on various parameters, the choice of which significantly affects… show more
PDF

Highlights - Most important sentences from the article

2016-06-14
1606.04414 | stat.ML

In many applications of black-box optimization, one can evaluate multiple points simultaneously, e.g… show more
PDF

Highlights - Most important sentences from the article

2018-11-23

Bayesian optimization usually assumes that a Bayesian prior is given. However, the strong theoretica… show more
PDF

Highlights - Most important sentences from the article

2015-11-24
1511.07916 | cs.CL

This is a lecture note for the course DS-GA 3001 <Natural Language Understanding with Distributed Re… show more
PDF

Highlights - Most important sentences from the article

2015-11-19

Sequences have become first class citizens in supervised learning thanks to the resurgence of recurr… show more
PDF

Highlights - Most important sentences from the article

2016-01-29

There has been a growing interest in using non-parametric regression methods like Gaussian Process (… show more
PDF

Highlights - Most important sentences from the article

2017-03-06
1703.01968 | stat.ML

Entropy Search (ES) and Predictive Entropy Search (PES) are popular and empirically successful Bayes… show more
PDF

Highlights - Most important sentences from the article

2015-10-21
1510.06423 | stat.ML

Recently, there has been rising interest in Bayesian optimization -- the optimization of an unknown … show more
PDF

Highlights - Most important sentences from the article

2019-02-08

Bayesian optimization is known to be difficult to scale to high dimensions, because the acquisition … show more
PDF

Highlights - Most important sentences from the article

2018-07-01
1807.00373 | stat.AP

Bayesian optimization has emerged as a strong candidate tool for global optimization of functions wi… show more
PDF

Highlights - Most important sentences from the article

2012-06-13

Machine learning algorithms frequently require careful tuning of model hyperparameters, regularizati… show more
PDF

Highlights - Most important sentences from the article

2018-11-25

This paper focuses on the problem of determining as large a region as possible where a function exce… show more
PDF

Highlights - Most important sentences from the article

2018-10-25

In this paper, we consider the problem of Gaussian process (GP) optimization with an added robustnes… show more
PDF