Search

In the reservoir computing literature, the information processing capacity is frequently used to characterize the computing capabilities of a reservoir. However, it remains unclear how the information processing capacity connects to the performance on specific tasks. We demonstrate on a set of standard benchmark tasks that the total information processing capacity correlates poorly with task specific performance. Further, we derive an expression for the normalized mean square error of a task as a weighted function of the individual information processing capacities. Mathematically, the derivation requires the task to have the same input distribution as used to calculate the information processing capacities. We test our method on a range of tasks that violate this requirement and find good qualitative agreement between the predicted and the actual errors as long as the task input sequences do not have long autocorrelation times. Our method offers deeper insight into the principles governing reservoir computing performance. It also increases the utility of the evaluation of information processing capacities, which are typically defined on i.i.d. input, even if specific tasks deliver inputs stemming from different distributions. Moreover, it offers the possibility of reducing the experimental cost of optimizing physical reservoirs, such as those implemented in photonic systems.

Delivery companies are affected by emission-sensitive traffic management systems. These systems are installed in cities to react instantly to emission hot-spots through adapting traffic light programs at intersections. This results in a change of the travel times for the delivery vehicle. In this paper, we model the problem in a dynamic vehicle routing problem with stochastic transition of deterministic travel time matrices. To solve this problem, we apply approximate value iteration, a method of approximate dynamic programming, to anticipate future travel time matrix changes in dynamic routing decisions. We vary the approach in the level of information about the state of the traffic management system. This allows to distinguish the required information for the routing decisions. Further, we compare the dynamic and anticipatory routing policies with a static a priori routing. Computational studies show an improved tour duration of routing with traffic management information over the a priori routing by up to 6.5%. We further show that an efficient representation of the traffic management system status in the approximate value iteration approach is mandatory to achieve sufficient anticipation.

In city logistics, courier express and parcel services (CEP) deal with last mile deliveries in an urban environment. CEP route vehicles to deliver parcels to costumers distributed in the cities area. Customers expect fast and reliable services to reasonable prices. As a result, CEP have to plan and execute their routing efficiently. In a city logistic environment, CEP faces several challenges. Especially, travel times between customers are not deterministic but uncertain and differ during the day regarding traffic volumes and stochastic events like congestion. Further, city traffic flows are controlled by traffic control management (TM). TM decisions additionally impact travel times for CEP. This paper shows the impact of TM decisions to travel times and emphasizes the benefit of cooperational planning between CEP and TM.