PyTorchKR Papers

최신 논문

총 1,869편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1707.04402

Lenient Multi-Agent Deep Reinforcement Learning

Much of the success of single agent deep reinforcement learning (DRL) in recent years can be attributed to the use of experience replay memories (ERM), which allow Deep Q-Networks (DQNs) to be trained efficiently through sampling stored state transitions. However, care is required when using ERMs for multi-agent deep reinforcement learning (MA-DRL), as stored transitions can become outdated because agents update their policies in parallel [11]. In this work we apply leniency [23] to MA-DRL. Lenient agents map state-action pairs to decaying temperature values that control the amount of leniency applied towards negative policy updates that are sampled from the ERM. This introduces optimism in the value-function update, and has been shown to facilitate cooperation in tabular fully-cooperative multi-agent reinforcement learning problems. We evaluate our Lenient-DQN (LDQN) empirically against the related Hysteretic-DQN (HDQN) algorithm [22] as well as a modified version we call scheduled-HDQN, that uses average reward learning near terminal states. Evaluations take place in extended variations of the Coordinated Multi-Agent Object Transportation Problem (CMOTP) [8] which include fully-cooperative sub-tasks and stochastic rewards. We find that LDQN agents are more likely to converge to the optimal policy in a stochastic reward CMOTP compared to standard and scheduled-HDQN agents.

arXiv:1803.02018

Intent-aware Multi-agent Reinforcement Learning

This paper proposes an intent-aware multi-agent planning framework as well as a learning algorithm. Under this framework, an agent plans in the goal space to maximize the expected utility. The planning process takes the belief of other agents' intents into consideration. Instead of formulating the learning problem as a partially observable Markov decision process (POMDP), we propose a simple but effective linear function approximation of the utility function. It is based on the observation that for humans, other people's intents will pose an influence on our utility for a goal. The proposed framework has several major advantages: i) it is computationally feasible and guaranteed to converge. ii) It can easily integrate existing intent prediction and low-level planning algorithms. iii) It does not suffer from sparse feedbacks in the action space. We experiment our algorithm in a real-world problem that is non-episodic, and the number of agents and goals can vary over time. Our algorithm is trained in a scene in which aerial robots and humans interact, and tested in a novel scene with a different environment. Experimental results show that our algorithm achieves the best performance and human-like behaviors emerge during the dynamic process.

arXiv:1803.03491

Valuing knowledge, information and agency in Multi-agent Reinforcement Learning: a case study in smart buildings

Increasing energy efficiency in buildings can reduce costs and emissions substantially. Historically, this has been treated as a local, or single-agent, optimization problem. However, many buildings utilize the same types of thermal equipment e.g. electric heaters and hot water vessels. During operation, occupants in these buildings interact with the equipment differently thereby driving them to diverse regions in the state-space. Reinforcement learning agents can learn from these interactions, recorded as sensor data, to optimize the overall energy efficiency. However, if these agents operate individually at a household level, they can not exploit the replicated structure in the problem. In this paper, we demonstrate that this problem can indeed benefit from multi-agent collaboration by making use of targeted exploration of the state-space allowing for better generalization. We also investigate trade-offs between integrating human knowledge and additional sensors. Results show that savings of over 40% are possible with collaborative multi-agent systems making use of either expert knowledge or additional sensors with no loss of occupant comfort. We find that such multi-agent systems comfortably outperform comparable single agent systems.

arXiv:1803.08884

Inequity aversion improves cooperation in intertemporal social dilemmas

Groups of humans are often able to find ways to cooperate with one another in complex, temporally extended social dilemmas. Models based on behavioral economics are only able to explain this phenomenon for unrealistic stateless matrix games. Recently, multi-agent reinforcement learning has been applied to generalize social dilemma problems to temporally and spatially extended Markov games. However, this has not yet generated an agent that learns to cooperate in social dilemmas as humans do. A key insight is that many, but not all, human individuals have inequity averse social preferences. This promotes a particular resolution of the matrix game social dilemma wherein inequity-averse individuals are personally pro-social and punish defectors. Here we extend this idea to Markov games and show that it promotes cooperation in several types of sequential social dilemma, via a profitable interaction with policy learnability. In particular, we find that inequity aversion improves temporal credit assignment for the important class of intertemporal social dilemmas. These results help explain how large-scale cooperation may emerge and persist.

arXiv:1802.09640

Modeling Others using Oneself in Multi-Agent Reinforcement Learning

We consider the multi-agent reinforcement learning setting with imperfect information in which each agent is trying to maximize its own utility. The reward function depends on the hidden state (or goal) of both agents, so the agents must infer the other players' hidden goals from their observed behavior in order to solve the tasks. We propose a new approach for learning in these domains: Self Other-Modeling (SOM), in which an agent uses its own policy to predict the other agent's actions and update its belief of their hidden state in an online manner. We evaluate this approach on three different tasks and show that the agents are able to learn better policies using their estimate of the other players' hidden states, in both cooperative and adversarial settings.

arXiv:1804.00146

Towards Learning Transferable Conversational Skills using Multi-dimensional Dialogue Modelling

Recent statistical approaches have improved the robustness and scalability of spoken dialogue systems. However, despite recent progress in domain adaptation, their reliance on in-domain data still limits their cross-domain scalability. In this paper, we argue that this problem can be addressed by extending current models to reflect and exploit the multi-dimensional nature of human dialogue. We present our multi-dimensional, statistical dialogue management framework, in which transferable conversational skills can be learnt by separating out domain-independent dimensions of communication and using multi-agent reinforcement learning. Our initial experiments with a simulated user show that we can speed up the learning process by transferring learnt policies.

arXiv:1804.03980

Emergent Communication through Negotiation

Multi-agent reinforcement learning offers a way to study how communication could emerge in communities of agents needing to solve specific problems. In this paper, we study the emergence of communication in the negotiation environment, a semi-cooperative model of agent interaction. We introduce two communication protocols -- one grounded in the semantics of the game, and one which is \textit{a priori} ungrounded and is a form of cheap talk. We show that self-interested agents can use the pre-grounded communication channel to negotiate fairly, but are unable to effectively use the ungrounded channel. However, prosocial agents do learn to use cheap talk to find an optimal negotiating strategy, suggesting that cooperation is necessary for language to emerge. We also study communication behaviour in a setting where one agent interacts with agents in a community with different levels of prosociality and show how agent identifiability can aid negotiation.

arXiv:1804.05464

On Gradient-Based Learning in Continuous Games

We formulate a general framework for competitive gradient-based learning that encompasses a wide breadth of multi-agent learning algorithms, and analyze the limiting behavior of competitive gradient-based learning algorithms using dynamical systems theory. For both general-sum and potential games, we characterize a non-negligible subset of the local Nash equilibria that will be avoided if each agent employs a gradient-based learning algorithm. We also shed light on the issue of convergence to non-Nash strategies in general- and zero-sum games, which may have no relevance to the underlying game, and arise solely due to the choice of algorithm. The existence and frequency of such strategies may explain some of the difficulties encountered when using gradient descent in zero-sum games as, e.g., in the training of generative adversarial networks. To reinforce the theoretical contributions, we provide empirical results that highlight the frequency of linear quadratic dynamic games (a benchmark for multi-agent reinforcement learning) that admit global Nash equilibria that are almost surely avoided by policy gradient.

arXiv:1709.04326공식 코드 · Papers with Code

Learning with Opponent-Learning Awareness

Multi-agent settings are quickly gathering importance in machine learning. This includes a plethora of recent work on deep multi-agent reinforcement learning, but also can be extended to hierarchical RL, generative adversarial networks and decentralised optimisation. In all these settings the presence of multiple learning agents renders the training problem non-stationary and often leads to unstable training or undesired final results. We present Learning with Opponent-Learning Awareness (LOLA), a method in which each agent shapes the anticipated learning of the other agents in the environment. The LOLA learning rule includes a term that accounts for the impact of one agent's policy on the anticipated parameter update of the other agents. Results show that the encounter of two LOLA agents leads to the emergence of tit-for-tat and therefore cooperation in the iterated prisoners' dilemma, while independent learning does not. In this domain, LOLA also receives higher payouts compared to a naive learner, and is robust against exploitation by higher order gradient-based methods. Applied to repeated matching pennies, LOLA agents converge to the Nash equilibrium. In a round robin tournament we show that LOLA agents successfully shape the learning of a range of multi-agent learning algorithms from literature, resulting in the highest average returns on the IPD. We also show that the LOLA update rule can be efficiently calculated using an extension of the policy gradient estimator, making the method suitable for model-free RL. The method thus scales to large parameter and input spaces and nonlinear function approximators. We apply LOLA to a grid world task with an embedded social dilemma using recurrent policies and opponent modelling. By explicitly considering the learning of the other agent, LOLA agents learn to cooperate out of self-interest. The code is at github.com/alshedivat/lola.

arXiv:1702.08887

Stabilising Experience Replay for Deep Multi-Agent Reinforcement Learning

Many real-world problems, such as network packet routing and urban traffic control, are naturally modeled as multi-agent reinforcement learning (RL) problems. However, existing multi-agent RL methods typically scale poorly in the problem size. Therefore, a key challenge is to translate the success of deep learning on single-agent RL to the multi-agent setting. A major stumbling block is that independent Q-learning, the most popular multi-agent RL method, introduces nonstationarity that makes it incompatible with the experience replay memory on which deep Q-learning relies. This paper proposes two methods that address this problem: 1) using a multi-agent variant of importance sampling to naturally decay obsolete data and 2) conditioning each agent's value function on a fingerprint that disambiguates the age of the data sampled from the replay memory. Results on a challenging decentralised variant of StarCraft unit micromanagement confirm that these methods enable the successful combination of experience replay with multi-agent RL.

arXiv:1805.08776

Scalable Centralized Deep Multi-Agent Reinforcement Learning via Policy Gradients

In this paper, we explore using deep reinforcement learning for problems with multiple agents. Most existing methods for deep multi-agent reinforcement learning consider only a small number of agents. When the number of agents increases, the dimensionality of the input and control spaces increase as well, and these methods do not scale well. To address this, we propose casting the multi-agent reinforcement learning problem as a distributed optimization problem. Our algorithm assumes that for multi-agent settings, policies of individual agents in a given population live close to each other in parameter space and can be approximated by a single policy. With this simple assumption, we show our algorithm to be extremely effective for reinforcement learning in multi-agent settings. We demonstrate its effectiveness against existing comparable approaches on co-operative and competitive tasks.

arXiv:1803.09928

Entropy based Independent Learning in Anonymous Multi-Agent Settings

Efficient sequential matching of supply and demand is a problem of interest in many online to offline services. For instance, Uber, Lyft, Grab for matching taxis to customers; Ubereats, Deliveroo, FoodPanda etc for matching restaurants to customers. In these online to offline service problems, individuals who are responsible for supply (e.g., taxi drivers, delivery bikes or delivery van drivers) earn more by being at the "right" place at the "right" time. We are interested in developing approaches that learn to guide individuals to be in the "right" place at the "right" time (to maximize revenue) in the presence of other similar "learning" individuals and only local aggregated observation of other agents states (e.g., only number of other taxis in same zone as current agent). A key characteristic of the domains of interest is that the interactions between individuals are anonymous, i.e., the outcome of an interaction (competing for demand) is dependent only on the number and not on the identity of the agents. We model these problems using the Anonymous MARL (AyMARL) model. The key contribution of this paper is in employing principle of maximum entropy to provide a general framework of independent learning that is both empirically effective (even with only local aggregated information of agent population distribution) and theoretically justified. Finally, our approaches provide a significant improvement with respect to joint and individual revenue on a generic simulator for online to offline services and a real world taxi problem over existing approaches. More importantly, this is achieved while having the least variance in revenues earned by the learning individuals, an indicator of fairness.

arXiv:1806.00877

Multi-Agent Reinforcement Learning via Double Averaging Primal-Dual Optimization

Despite the success of single-agent reinforcement learning, multi-agent reinforcement learning (MARL) remains challenging due to complex interactions between agents. Motivated by decentralized applications such as sensor networks, swarm robotics, and power grids, we study policy evaluation in MARL, where agents with jointly observed state-action pairs and private local rewards collaborate to learn the value of a given policy. In this paper, we propose a double averaging scheme, where each agent iteratively performs averaging over both space and time to incorporate neighboring gradient information and local reward information, respectively. We prove that the proposed algorithm converges to the optimal solution at a global geometric rate. In particular, such an algorithm is built upon a primal-dual reformulation of the mean squared projected Bellman error minimization problem, which gives rise to a decentralized convex-concave saddle-point problem. To the best of our knowledge, the proposed double averaging primal-dual optimization algorithm is the first to achieve fast finite-time convergence on decentralized convex-concave saddle-point problems.

arXiv:1803.11485공식 코드 · Papers with Code

QMIX: Monotonic Value Function Factorisation for Deep Multi-Agent Reinforcement Learning

In many real-world settings, a team of agents must coordinate their behaviour while acting in a decentralised way. At the same time, it is often possible to train the agents in a centralised fashion in a simulated or laboratory setting, where global state information is available and communication constraints are lifted. Learning joint action-values conditioned on extra state information is an attractive way to exploit centralised learning, but the best strategy for then extracting decentralised policies is unclear. Our solution is QMIX, a novel value-based method that can train decentralised policies in a centralised end-to-end fashion. QMIX employs a network that estimates joint action-values as a complex non-linear combination of per-agent values that condition only on local observations. We structurally enforce that the joint-action value is monotonic in the per-agent values, which allows tractable maximisation of the joint action-value in off-policy learning, and guarantees consistency between the centralised and decentralised policies. We evaluate QMIX on a challenging set of StarCraft II micromanagement tasks, and show that QMIX significantly outperforms existing value-based multi-agent reinforcement learning methods.

arXiv:1802.05438공식 코드 · Papers with Code

Mean Field Multi-Agent Reinforcement Learning

Existing multi-agent reinforcement learning methods are limited typically to a small number of agents. When the agent number increases largely, the learning becomes intractable due to the curse of the dimensionality and the exponential growth of agent interactions. In this paper, we present \emph{Mean Field Reinforcement Learning} where the interactions within the population of agents are approximated by those between a single agent and the average effect from the overall population or neighboring agents; the interplay between the two entities is mutually reinforced: the learning of the individual agent's optimal policy depends on the dynamics of the population, while the dynamics of the population change according to the collective patterns of the individual policies. We develop practical mean field Q-learning and mean field Actor-Critic algorithms and analyze the convergence of the solution to Nash equilibrium. Experiments on Gaussian squeeze, Ising model, and battle games justify the learning effectiveness of our mean field approaches. In addition, we report the first result to solve the Ising model via model-free reinforcement learning methods.

arXiv:2610.15969v1cs.AI

PopGen: 제로샷 조정을 위한 의미론적으로 다양한 개체군 생성

PopGen: Generating Semantically Diverse Populations for Zero-Shot Coordination

다중 에이전트 강화학습(multi-agent reinforcement learning)에서 학습(training) 중 협력자에게 다양한 전략을 가진 파트너를 노출하는 것은 테스트 시 더 나은 협업을 달성하기 위한 핵심 요소입니다. 그러나 기존 방법들은 일반적으로 다양성을 측정하기 위해 궤적 수준(trajectory-level)의 대리 지표에 의존하므로, 동일한 전략에서 나온 서로 다른 두 궤적을 다양하다고 잘못 분류할 수 있습니다. 대신 본 연구의 핵심 통찰은 인간이 고수준의 전략을 인식할 수 있는 것과 유사하게, 환경(environment) 내에서 행동적으로 뚜렷이 구별되는 에이전트(agent)를 서술하는 대규모 언어 모델(LLM)의 능력을 활용하는 것입니다. 이를 위해 본 연구에서는 LLM을 사용하여 처음부터 의미론적으로 다양한 에이전트를 합성하는 프레임워크인 PopGen을 소개합니다. PopGen은 먼저 에이전트 행동에 대한 다양한 텍스트 서술 세트를 생성한 다음, 각 행동을 환경 내에서 실행할 수 있는 코드 정책(code policy)으로 변환합니다. 본 연구는 PopGen 개체군을 대상으로 학습된 협력자가 처음 보는 파트너와도 견고한 제로샷(zero-shot) 협력을 달성하며, 네 가지 협업 환경 전반에서 개체군 기반 학습(population-play) 베이스라인들을 능가함을 보입니다. 나아가 의미론적 다양성이 PopGen 개체군 크기에 따라 확장되어 협력자의 최악의 경우(worst-case) 성능을 향상시킴을 보이고, 다양성을 극대화하면서도 개체군 품질을 유지하는 설계 선택 사항들을 식별합니다. 연구 코드는 https://github.com/labicon/PopGen 에서 확인할 수 있습니다.

arXiv:2610.15932v1cs.LG

DiffNash: 확산 정책을 활용한 고차원 연속 행동 공간에서의 내시 균형 탐색

DiffNash: Finding Nash Equilibrium in High-Dimensional Continuous Action Spaces with Diffusion Policies

고차원 연속 게임에서 균형(equilibrium)을 계산하려면 풍부한 다봉(multi-modal) 전략 분포를 학습하고 표현해야 하지만, 기존의 신경망 정책으로는 이를 달성하기 어렵습니다. 확산 모델(diffusion model)은 자연스러운 해결책을 제공하지만, 표준 확산 학습(training)은 고정된 목표 분포로부터의 직접적인 샘플을 필요로 하며, 이는 반복적인 게임 해결 업데이트 과정에서는 일반적으로 얻기 어렵습니다. 본 논문에서는 온라인 미러 하강법(OMD) 및 가상 플레이(FP)의 확산 기반 변형을 가능하게 하는 균형 계산용 확산 기반 프레임워크를 제안합니다. 제안하는 접근법은 각 업데이트된 정책으로부터 직접 샘플을 추출할 필요 없이 확산 정책을 학습시키는 재가중(reweighted) 스코어 매칭 목적식을 유도합니다. 본 연구의 핵심 방법론인 DiffNash는 리플레이 버퍼를 통해 과거 행동을 재사용함으로써 OMD의 빠른 수렴(convergence)과 FP의 샘플 효율성을 결합합니다. 샘플 재사용으로 인해 발생하는 분포 변화(distribution shift)를 보정하기 위해, 확산 우도(likelihood)를 직접 평가하지 않고도 해당 중요도 가중치(weight)에 대한 다루기 쉬운 대리 지표를 제공하는, 현재 및 과거 정책의 증거 하한(ELBO) 차이에 대한 비편향 추정량을 유도합니다. 산림 보호 게임 및 연속 블로토 대령(Colonel Blotto) 게임에서 제안 방식을 평가한 결과, 최소한의 하이퍼파라미터(hyperparameter) 튜닝만으로 가장 강력한 기준선(baseline) 대비 내시 격차(Nash gap)를 각각 35.6% 및 24.7% 감소시켰습니다.

arXiv:2610.13737v1cs.LG

SafeDiSTL: 안전한 다중 에이전트 강화학습을 위한 공동 제약조건 분해

SafeDiSTL: Decomposing Joint Constraints for Safe Multi-Agent Reinforcement Learning

협력적 다중 에이전트 태스크는 종종 공유 목표나 집합적 안전성(collective safety)에 대한 제약조건을 부과합니다. 팀 수준의 위반 비용 함수는 이러한 제약조건 결과를 측정할 수 있지만, 어떤 에이전트의 행동이 해당 위반에 가장 큰 영향을 미치는지 구분하지 못합니다. 본 논문에서는 결합 제약조건 측정 하에서도 신용 할당(credit assignment)이 가능한 제약 다중 에이전트 강화학습(MARL) 프레임워크인 SafeDiSTL을 제안합니다. 우리는 신호 시간 논리(Signal Temporal Logic, STL) 강건성을 활용하여 실행 최댓값(running-maximum) 비용 증분을 통해 에피소드 수준의 위반을 측정합니다. 매끄러운 전체 에피소드 강건성과 미분 가능한 동역학은 행동 민감도(action sensitivity)를 제공하며, 이는 동일한 결정 시점에 각 비용 증분을 에이전트 전반에 걸쳐 분배합니다. 이러한 민감도는 각 비용 증분에서 에이전트의 기여도를 가중할 때 지연된 영향을 고려합니다. 우리는 협력적 화물 운송, 연결 항법, 네트워크 트래픽 제어 태스크에서 본 프레임워크를 연구합니다. 제약 baselines과 비교했을 때, SafeDiSTL은 운송 태스크에서 안전 성공률을 1.6~5.1배 달성하고, 항법 안전 성공률을 11.1% 향상시키며, 트래픽 태스크에서 50~68% 더 높은 처리량을 보입니다. 실험 결과는 학습된 정책이 제약 없는 정책의 성능을 유지하면서도 공유 제약조건을 다루는 데 있어 크게 향상되었음을 입증합니다.

arXiv:2610.13534v1cs.AI

MAGMA-Bench: 동적 태스크 및 실행 실패 환경에서의 장기 임바디드 에이전트 평가

MAGMA-Bench: Evaluating Long-Horizon Embodied Agents under Dynamic Tasks and Execution Failures

장기(long-horizon) 임바디드 에이전트(embodied agent)는 실행이 진행됨에 따라 태스크 관련 정보를 유지하고, 이종 로봇들을 조율하며, 의사결정을 적응적으로 변경할 수 있어야 합니다. 기존 벤치마크는 동적 태스크 실행 및 임바디드 멀티 에이전트 조율을 조사하지만, 둘의 교차점에 대한 평가는 제한적입니다. 본 연구에서는 동적 멀티 로봇 환경에서의 지속적 태스크 추론을 위한 아키텍처 비종속적 벤치마크인 MAGMA-Bench를 소개합니다. MAGMA-Bench는 이종 역량을 가진 로봇 간의 조율이 필요한 시나리오와 함께, 실행 실패, 태스크 중단, 미션 업데이트를 포함한 통제된 온라인 섭동(online perturbation) 및 늘어나는 태스크 범위에 걸쳐 에이전트를 평가합니다. 본 벤치마크는 구조화된 스킬 수준 인터페이스를 활용하여 태스크 수준 추론을 인지 및 저수준 제어로부터 분리합니다. 저자들은 로컬 배포가 가능한 오픈 가중치 플래너, 멀티 에이전트 상태 추적 전략, 그리고 프론티어 모델들을 평가합니다. 평가된 가장 강력한 모델조차도 섭동이 없는 에피소드의 상당 부분에서 실패하며, 온라인 섭동 하에서는 성공률이 더욱 감소합니다. 지도 적응(supervised adaptation)은 소형 모델의 성능을 크게 향상시키지만, 더 길고 다중 로봇이 관여하는 미션에서는 성능이 여전히 크게 낮게 유지됩니다. 저자들은 벤치마크, 261,646개의 태스크 수준 학습 예제, 그리고 아키텍처 비종속적 전문가 궤적 생성 파이프라인을 공개합니다.

arXiv:2610.13439v1cs.MA

자율주행차의 집합적 경로 선택에는 인간 운전자와의 공존을 위한 사회적 인식이 필요하다

Collective Autonomous Vehicle Routing Needs Social Awareness for Coexistence with Human Drivers

실제 교통 네트워크에서 자율주행차(AV)의 집합적 경로 선택은 인간 운전 차량(HDV)에 외부효과를 유발할 수 있는데, AV에게 유리한 경로가 다른 도로 이용자의 통행 시간을 증가시킬 수 있기 때문이다. 이를 해결하기 위해 우리는 이기적인 통행 시간 기반 AV 보상에 사회적 요소인 한계 비용(marginal cost)을 도입한다. 한계 비용은 해당 AV를 제거한 반사실적(counterfactual) 시뮬레이션에서의 통행 시간과 비교하여, 해당 AV가 다른 도로 이용자에게 미치는 영향을 포착한다. 우리는 다중 에이전트 강화학습(Multi-Agent Reinforcement Learning, MARL)을 사용하여 혼합 자율주행(mixed-autonomy) 교통 환경에서 AV의 경로 선택을 연구하며, 여기서 AV는 교통 환경과의 반복적인 상호작용을 통해 경로 선택 정책을 학습한다. Ingolstadt와 Provins의 실제 네트워크에 대한 실험 결과, 한계 비용을 AV 보상에 포함하면 이기적인 AV 경로 선택에 비해 HDV 통행 시간 증가를 제한하고 AV 통행 시간을 단축할 뿐만 아니라, 전체 시스템 성능도 향상되는 것으로 나타났다. Ingolstadt에서 AV 40%의 이기적인 경로 선택은 다른 차량의 통행 시간을 1.7% 증가시키지만, 사회적 보상을 포함하면 이 수치는 1.1%로 감소한다. 흥미롭게도 전체 시스템 성능이 향상될 뿐만 아니라, AV 자신도 사회적으로 행동할 때 더 나은 결과를 얻는다. 이러한 효과는 실험 대상 도시, 보급률(penetration rate), 반복 실험 전반에 걸쳐 일관되게 나타나지만, 보상에서 사회적 요소의 가중치가 증가하면 시스템이 원하는 성능에 도달하지 못한다. 혼합 자율주행 교통에서 사회적 AV 경로 선택을 보다 확장 가능하게 만들기 위해, 우리는 교통 시스템에 가장 큰 영향을 미치는 경로 선택을 하는 AV에 대해서만 한계 비용을 계산한다. 이를 통해 반사실적 시뮬레이션의 횟수를 줄이면서도 훨씬 낮은 계산 비용으로 유사한 성능을 달성한다. 우리의 결과는 AV 경로 선택이 이기적인 보상을 넘어, 모든 교통 참여자의 이익을 위해 목적 함수에 사회적 요소를 포함해야 함을 시사한다.