In multi-objective problems, it is key to find compromising solutions that balance different objectives. The linear scalarization function is often utilized to translate the multi-objective nature of a problem into a standard, single-objective problem. Generally, it is noted that such as linear combination can only find solutions in convex areas of the Pareto front, therefore making the method inapplicable in situations where the shape of the front is not known beforehand, as is often the case. We propose a non-linear scalarization function, called the Chebyshev scalarization function, as a basis for action selection strategies in multi-objective reinforcement learning. The Chebyshev scalarization method overcomes the flaws of the linear scalarization function as it can (i) discover Pareto optimal solutions regardless of the shape of the front, i.e. convex as well as non-convex , (ii) obtain a better spread amongst the set of Pareto optimal solutions and (iii) is not particularly dependent on the actual weights used.
1 February 2019 in Swarm and Evolutionary Computation
Balancing the ratio between exploration and exploitation is an important problem in reinforcement learning. This paper evaluates four different exploration strategies combined with Q-learning using random stochastic mazes to investigate their performances. We will compare: UCB-1, softmax, ∈-greedy, and pursuit. For this purpose we adapted the UCB-1 and pursuit strategies to be used in the Q-learning algorithm. The mazes consist of a single optimal goal state and two suboptimal goal states that lie closer to the starting position of the agent, which makes efficient exploration an important part of the learning agent. Furthermore, we evaluate two different kinds of reward functions, a normalized one with rewards between 0 and 1, and an unnormalized reward function that penalizes the agent for each step with a negative reward. We have performed an extensive grid-search to find the best parameters for each method and used the best parameters on novel randomly generated maze problems of different sizes. The results show that softmax exploration outperforms the other strategies, although it is harder to tune its temperature parameter. The worst performing exploration strategy is ∈-greedy.
We propose an algorithmic framework for multi-objective multi-armed bandits with multiple rewards. Different partial order relationships from multi-objective optimization can be considered for a set of reward vectors, such as scalarization functions and Pareto search. A scalarization function transforms the multi-objective environment into a single objective environment and are a popular choice in multi-objective reinforcement learning. Scalarization techniques can be straightforwardly implemented into the current multi-armed bandit framework, but the efficiency of these algorithms depends very much on their type, linear or non-linear (e.g. Chebyshev), and their parameters. Using Pareto dominance order relationship allows to explore the multi-objective environment directly, however this can result in large sets of Pareto optimal solutions. In this paper we propose and evaluate the performance of multi-objective MABs using three regret metric criteria. The standard UCB1 is extended to scalarized multi-objective UCB1 and we propose a Pareto UCB1 algorithm. Both algorithms are proven to have a logarithmic upper bound for their expected regret. We also introduce a variant of the scalarized multi-objective UCB1 that removes online inefficient scalarizations in order to improve the algorithm's efficiency. These algorithms are experimentally compared on multi-objective Bernoulli distributions, Pareto UCB1 being the algorithm with the best empirical performance.