اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدOn Multi-objective Policy Optimization as a Tool for Reinforcement Learning
102
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Many advances that have improved the robustness and efficiency of deep reinforcement learning (RL) algorithms can, in one way or another, be understood as introducing additional objectives, or constraints, in the policy optimization step. This includes ideas as far ranging as exploration bonuses, entropy regularization, and regularization toward teachers or data priors when learning from experts or in offline RL. Often, task reward and auxiliary objectives are in conflict with each other and it is therefore natural to treat these examples as instances of multi-objective (MO) optimization problems. We study the principles underlying MORL and introduce a new algorithm, Distillation of a Mixture of Experts (DiME), that is intuitive and scale-invariant under some conditions. We highlight its strengths on standard MO benchmark problems and consider case studies in which we recast offline RL and learning from experts as MO problems. This leads to a natural algorithmic formulation that sheds light on the connection between existing approaches. For offline RL, we use the MO perspective to derive a simple algorithm, that optimizes for the standard RL objective plus a behavioral cloning term. This outperforms state-of-the-art on two established offline RL benchmarks.
قيم البحث
اقرأ أيضاً
Many engineering problems have multiple objectives, and the overall aim is to optimize a non-linear function of these objectives. In this paper, we formulate the problem of maximizing a non-linear concave function of multiple long-term objectives. A
policy-gradient based model-free algorithm is proposed for the problem. To compute an estimate of the gradient, a biased estimator is proposed. The proposed algorithm is shown to achieve convergence to within an $epsilon$ of the global optima after sampling $mathcal{O}(frac{M^4sigma^2}{(1-gamma)^8epsilon^4})$ trajectories where $gamma$ is the discount factor and $M$ is the number of the agents, thus achieving the same dependence on $epsilon$ as the policy gradient algorithm for the standard reinforcement learning.
Reinforcement learning agents need exploratory behaviors to escape from local optima. These behaviors may include both immediate dithering perturbation and temporally consistent exploration. To achieve these, a stochastic policy model that is inheren
tly consistent through a period of time is in desire, especially for tasks with either sparse rewards or long term information. In this work, we introduce a novel on-policy temporally consistent exploration strategy - Neural Adaptive Dropout Policy Exploration (NADPEx) - for deep reinforcement learning agents. Modeled as a global random variable for conditional distribution, dropout is incorporated to reinforcement learning policies, equipping them with inherent temporal consistency, even when the reward signals are sparse. Two factors, gradients alignment with the objective and KL constraint in policy space, are discussed to guarantee NADPEx policys stable improvement. Our experiments demonstrate that NADPEx solves tasks with sparse reward while naive exploration and parameter noise fail. It yields as well or even faster convergence in the standard mujoco benchmark for continuous control.
Multi-objective optimization (MOO) is a prevalent challenge for Deep Learning, however, there exists no scalable MOO solution for truly deep neural networks. Prior work either demand optimizing a new network for every point on the Pareto front, or in
duce a large overhead to the number of trainable parameters by using hyper-networks conditioned on modifiable preferences. In this paper, we propose to condition the network directly on these preferences by augmenting them to the feature space. Furthermore, we ensure a well-spread Pareto front by penalizing the solutions to maintain a small angle to the preference vector. In a series of experiments, we demonstrate that our Pareto fronts achieve state-of-the-art quality despite being computed significantly faster. Furthermore, we showcase the scalability as our method approximates the full Pareto front on the CelebA dataset with an EfficientNet network at a tiny training time overhead of 7% compared to a simple single-objective optimization. We make our code publicly available at https://github.com/ruchtem/cosmos.
Although deep reinforcement learning (deep RL) methods have lots of strengths that are favorable if applied to autonomous driving, real deep RL applications in autonomous driving have been slowed down by the modeling gap between the source (training)
domain and the target (deployment) domain. Unlike current policy transfer approaches, which generally limit to the usage of uninterpretable neural network representations as the transferred features, we propose to transfer concrete kinematic quantities in autonomous driving. The proposed robust-control-based (RC) generic transfer architecture, which we call RL-RC, incorporates a transferable hierarchical RL trajectory planner and a robust tracking controller based on disturbance observer (DOB). The deep RL policies trained with known nominal dynamics model are transfered directly to the target domain, DOB-based robust tracking control is applied to tackle the modeling gap including the vehicle dynamics errors and the external disturbances such as side forces. We provide simulations validating the capability of the proposed method to achieve zero-shot transfer across multiple driving scenarios such as lane keeping, lane changing and obstacle avoidance.
Federated learning has emerged as a promising, massively distributed way to train a joint deep model over large amounts of edge devices while keeping private user data strictly on device. In this work, motivated from ensuring fairness among users and
robustness against malicious adversaries, we formulate federated learning as multi-objective optimization and propose a new algorithm FedMGDA+ that is guaranteed to converge to Pareto stationary solutions. FedMGDA+ is simple to implement, has fewer hyperparameters to tune, and refrains from sacrificing the performance of any participating user. We establish the convergence properties of FedMGDA+ and point out its connections to existing approaches. Extensive experiments on a variety of datasets confirm that FedMGDA+ compares favorably against state-of-the-art.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
