Bus transit systems are the backbone of public transportation in the United States. An important indicator of the quality of service in such infrastructures is on-time performance at stops, with published transit schedules playing an integral role governing the level of success of the service. However there are relatively few optimization architectures leveraging stochastic search that focus on optimizing bus timetables with the objective of maximizing probability of bus arrivals at timepoints with delays within desired on-time ranges. In addition to this, there is a lack of substantial research considering monthly and seasonal variations of delay patterns integrated with such optimization strategies. To address these, this paper makes the following contributions to the corpus of studies on transit on-time performance optimization: (a) an unsupervised clustering mechanism is presented which groups months with similar seasonal delay patterns, (b) the problem is formulated as a single-objective optimization task and a greedy algorithm, a genetic algorithm (GA) as well as a particle swarm optimization (PSO) algorithm are employed to solve it, (c) a detailed discussion on empirical results comparing the algorithms are provided and sensitivity analysis on hyper-parameters of the heuristics are presented along with execution times, which will help practitioners looking at similar problems. The analyses conducted are insightful in the local context of improving public transit scheduling in the Nashville metro region as well as informative from a global perspective as an elaborate case study which builds upon the growing corpus of empirical studies using nature-inspired approaches to transit schedule optimization. Keywords: timetable optimization · genetic algorithm · particle swarm optimization · sensitivity analysis · scheduling 1 Introduction Bus systems are the backbone of public transportation in the US, carrying over 47% of all public passenger trips and 19,380 million passenger miles in the US [18] . For the majority of cities in the US which do not have enough urban forms or budget to build expensive transit infrastructures like subways, the reliance is on buses as the most important transit system since bus systems have advantages arXiv:1912.02574v1

Context plays a significant role in the generation of motion for dynamic agents in interactive environments. This work proposes a modular method that utilises a model of the environment to aid motion prediction of tracked agents. This paper shows that modelling the spatial and dynamic aspects of a given environment alongside the local per agent behaviour results in more accurate and informed long-term motion prediction. Further, we observe that this decoupling of dynamics and environment models allows for better generalisation to unseen environments, requiring that only a spatial representation of a new environment be learned. We highlight the model's prediction capability using a benchmark pedestrian tracking problem and by tracking a robot arm performing a tabletop manipulation task. The proposed approach allows for robust and data efficient forward modelling, and relaxes the need for full model re-training in new environments. We evaluate this through an ablation study which shows better performance gain when decoupling representation modules in addition to improved generalisation on tasks with dynamics unseen at training time.

Gaussian Processes (GPs) are widely employed in control and learning because of their principled treatment of uncertainty. However, tracking uncertainty for iterative, multi-step predictions in general leads to an analytically intractable problem. While approximation methods exist, they do not come with guarantees, making it difficult to estimate their reliability and to trust their predictions. In this work, we derive formal probability error bounds for iterative prediction and planning with GPs. Building on GP properties, we bound the probability that random trajectories lie in specific regions around the predicted values. Namely, given a tolerance $\epsilon > 0 $, we compute regions around the predicted trajectory values, such that GP trajectories are guaranteed to lie inside them with probability at least $1-\epsilon$. We verify experimentally that our method tracks the predictive uncertainty correctly, even when current approximation techniques fail. Furthermore, we show how the proposed bounds can be employed within a safe reinforcement learning framework to verify the safety of candidate control policies, guiding the synthesis of provably safe controllers.

In many collective decision making situations, agents vote to choose an alternative that best represents the preferences of the group. Agents may manipulate the vote to achieve a better outcome by voting in a way that does not reflect their true preferences. In real world voting scenarios, people often do not have complete information about other voter preferences and it can be computationally complex to identify a strategy that will maximize their expected utility. In such situations, it is often assumed that voters will vote truthfully rather than expending the effort to strategize. However, being truthful is just one possible heuristic that may be used. In this paper, we examine the effectiveness of heuristics in single winner and multi-winner approval voting scenarios with missing votes. In particular, we look at heuristics where a voter ignores information about other voting profiles and makes their decisions based solely on how much they like each candidate. In a behavioral experiment, we show that people vote truthfully in some situations and prioritize high utility candidates in others. We examine when these behaviors maximize expected utility and show how the structure of the voting environment affects both how well each heuristic performs and how humans employ these heuristics.

In this paper we propose the first machine teaching algorithm for multiple inverse reinforcement learners. Specifically, our contributions are: (i) we formally introduce the problem of teaching a sequential task to a heterogeneous group of learners; (ii) we identify conditions under which it is possible to conduct such teaching using the same demonstration for all learners; and (iii) we propose and evaluate a simple algorithm that computes a demonstration(s) ensuring that all agents in a heterogeneous class learn a task description that is compatible with the target task. Our analysis shows that, contrary to other teaching problems, teaching a heterogeneous class with a single demonstration may not be possible as the differences between agents increase. We also showcase the advantages of our proposed machine teaching approach against several possible alternatives.

The global autonomous agents market is expected to grow at a CAGR of 55.15% during the forecast period, 2019–2026. The pervasiveness of artificial intelligence is inciting the demand for autonomous agents across the globe. An autonomous agent, also known as an intelligent agent, operates and is enabled to act and react in a dynamic working environment. These agents could work independently or collaboratively, providing solutions to complex problems. Autonomous agents range from smartphone applications to autonomous robots such as shopping bots, personal agents, monitoring and surveillance agents, and data mining agents, among others.

We propose a novel approach to address one aspect of the non-stationarity problem in multi-agent reinforcement learning (RL), where the other agents may alter their policies due to environment changes during execution. This violates the Markov assumption that governs most single-agent RL methods and is one of the key challenges in multi-agent RL. To tackle this, we propose to train multiple policies for each agent and postpone the selection of the best policy at execution time. Specifically, we model the environment non-stationarity with a finite set of scenarios and train policies fitting each scenario. In addition to multiple policies, each agent also learns a policy predictor to determine which policy is the best with its local information. By doing so, each agent is able to adapt its policy when the environment changes and consequentially the other agents alter their policies during execution. We empirically evaluated our method on a variety of common benchmark problems proposed for multi-agent deep RL in the literature. Our experimental results show that the agents trained by our algorithm have better adaptiveness in changing environments and outperform the state-of-the-art methods in all the tested environments.

In this paper, we investigate learning temporal abstractions in cooperative multi-agent systems using the options framework (Sutton et al, 1999) and provide a model-free algorithm for this problem. First, we address the planning problem for the decentralized POMDP represented by the multi-agent system, by introducing a common information approach. We use common beliefs and broadcasting to solve an equivalent centralized POMDP problem. Then, we propose the Distributed Option Critic (DOC) algorithm, motivated by the work of Bacon et al (2017) in the single-agent setting. Our approach uses centralized option evaluation and decentralized intra-option improvement. We analyze theoretically the asymptotic convergence of DOC and validate its performance in grid-world environments, where we implement DOC using a deep neural network. Our experiments show that DOC performs competitively with state-of-the-art algorithms and that it is scalable when the number of agents increases.

--With the rapid evolution of wireless mobile devices, it emerges stronger incentive to design proper collaboration mechanisms among the intelligent agents. Following their individual observations, multiple intelligent agents could cooperate and gradually approach the final collective objective through continuously learning from the environment. In that regard, independent reinforcement learning (IRL) is often deployed within the multi-agent collaboration to alleviate the dilemma of non-stationary learning environment. However, behavioral strategies of the intelligent agents in IRL could only be formulated upon their local individual observations of the global environment, and appropriate communication mechanisms must be introduced to reduce their behavioral localities. In this paper, we tackle the communication problem among the intelligent agents in IRL by jointly adopting two mechanisms with different scales. For the large scale, we introduce the stigmergy mechanism as an indirect communication bridge among the independent learning agents and carefully design a mathematical representation to indicate the impact of digital pheromone. For the small scale, we propose a conflict-avoidance mechanism between adjacent agents by implementing an additionally embedded neural network to provide more opportunities for participants with higher action priorities. Besides, we also present a federal training method to effectively optimize the neural networks within each agent in a decentralized manner . Finally, we establish a simulation scenario where a number of mobile agents in a certain area move automatically to form a specified target shape, and demonstrate the superiorities of our proposed methods through extensive simulations. I NTRODUCTION With the rapid development of mobile wireless communication and IoTs (Internet of Things) technologies, many scenarios gradually arise where the collaboration among the involved intelligent agents is highly required, such as the deployment of unmanned aerial vehicles (UA Vs) [1]-[3], the distributed control in the field of industry automation [4]-[6], and mobile crowd sensing and computing (MCSC) [7], [8]. In these scenarios, traditional centralized control methods are usually impracticable because of the restriction from limited computing resources as well as the demand for ultra-low latency and ultra-high reliability. As an alternative, multi-agent collaboration can be introduced into these scenarios to reduce the pressure at the central controller side. As one of the primary goals in the field of artificial intelligence (AI), assisting autonomous agents to act optimally through the "trial-and-error" interaction process with the expected environment is regarded as an important target of reinforcement learning (RL) [9]-[11].

Artificial Intelligence and Machine Learning are increasingly seen as key technologies for buildin g more decentralised and resilient energy grids, but researchers must consider the ethical and social implications of their use E nergy grids are undergoing rapid changes, requiring new ways both to process the large amounts of data generated from the power system, but also - increasingly - to take smart operational decisions [1]. On the data side, the UK and most EU countries have committed to a target of offering a smart meter to every home by 2020 [ 2 ], with similar monitoring being installed in other parts of the energy network. This has led to some to refer to a "data tsunami", requiri ng development of new machine learning techniques to deal with the e nsuing challenge of extracting useful information from this data - often in real time. Another trend is the use of AI techniques (such as those from multi - agent systems, computational gam e theory and decision making under uncertainty) to take autonomous allocation and control decisions. This is driven increasingly by the moves towards more decentralised energy systems, where prosumers (consumers with own micro - generation and storage) can g enerate and source their own electricity through peer - to - peer (P2P) trading in local energy markets and community energy schemes.