In percutaneous intervention for treatment of coronary plaques, guidewire navigation is a primary procedure for stent delivery. Steering a flexible guidewire within coronary arteries requires considerable training, and the non-linearity between the control operation and the movement of the guidewire makes precise manipulation difficult. Here, we introduce a deep reinforcement learning(RL) framework for autonomous guidewire navigation in a robot-assisted coronary intervention. Using Rainbow, a segment-wise learning approach is applied to determine how best to accelerate training using human demonstrations with deep Q-learning from demonstrations (DQfD), transfer learning, and weight initialization. `State' for RL is customized as a focus window near the guidewire tip, and subgoals are placed to mitigate a sparse reward problem. The RL agent improves performance, eventually enabling the guidewire to reach all valid targets in `stable' phase. Our framework opens anew direction in the automation of robot-assisted intervention, providing guidance on RL in physical spaces involving mechanical fatigue.
In this paper, we propose a reinforcement learning based algorithm for rate-profile construction of Arikan's Polarization Assisted Convolutional (PAC) codes. This method can be used for any blocklength, rate, list size under successive cancellation list (SCL) decoding and convolutional precoding polynomial. To the best of our knowledge, we present, for the first time, a set of new reward and update strategies which help the reinforcement learning agent discover much better rate-profiles than those present in existing literature. Simulation results show that PAC codes constructed with the proposed algorithm perform better in terms of frame erasure rate (FER) compared to the PAC codes constructed with contemporary rate profiling designs for various list lengths. Further, by using a (64, 32) PAC code as an example, it is shown that the choice of convolutional precoding polynomial can have a significant impact on rate-profile construction of PAC codes.
Offline reinforcement learning (offline RL), which aims to find an optimal policy from a previously collected static dataset, bears algorithmic difficulties due to function approximation errors from out-of-distribution (OOD) data points. To this end, offline RL algorithms adopt either a constraint or a penalty term that explicitly guides the policy to stay close to the given dataset. However, prior methods typically require accurate estimation of the behavior policy or sampling from OOD data points, which themselves can be a non-trivial problem. Moreover, these methods under-utilize the generalization ability of deep neural networks and often fall into suboptimal solutions too close to the given dataset. In this work, we propose an uncertainty-based offline RL method that takes into account the confidence of the Q-value prediction and does not require any estimation or sampling of the data distribution. We show that the clipped Q-learning, a technique widely used in online RL, can be leveraged to successfully penalize OOD data points with high prediction uncertainties. Surprisingly, we find that it is possible to substantially outperform existing offline RL methods on various tasks by simply increasing the number of Q-networks along with the clipped Q-learning. Based on this observation, we propose an ensemble-diversified actor-critic algorithm that reduces the number of required ensemble networks down to a tenth compared to the naive ensemble while achieving state-of-the-art performance on most of the D4RL benchmarks considered.
The widespread diffusion of mobile phones is triggering an exponential growth of mobile data traffic that is likely to cause, in the near future, considerable traffic overload issues even in last-generation cellular networks. Offloading part of the traffic to other networks is considered a very promising approach and, in particular, in this paper, we consider offloading through opportunistic networks of users' devices. However, the performance of this solution strongly depends on the pattern of encounters between mobile nodes, which should therefore be taken into account when designing offloading control algorithms. In this paper, we propose an adaptive offloading solution based on the Reinforcement Learning framework and we evaluate and compare the performance of two well-known learning algorithms: Actor-Critic and Q-Learning. More precisely, in our solution the controller of the dissemination process, once trained, is able to select a proper number of content replicas to be injected into the opportunistic network to guarantee the timely delivery of contents to all interested users. We show that our system based on Reinforcement Learning is able to automatically learn a very efficient strategy to reduce the traffic on the cellular network, without relying on any additional context information about the opportunistic network. Our solution achieves a higher level of offloading with respect to other state-of-the-art approaches, in a range of different mobility settings. Moreover, we show that a more refined learning solution, based on the Actor-Critic algorithm, is significantly more efficient than a simpler solution based on Q-learning.
We propose a learning-based approach for the sparse Gaussian Elimination. There are many hard combinatorial optimization problems in modern sparse solver. These NP-hard problems could be handled in the framework of Markov Decision Process, especially the Q-Learning technique. We proposed some Q-Learning algorithms for the main modules of sparse solver: minimum degree ordering, task scheduling and adaptive pivoting. Finally, we recast the sparse solver into the framework of Q-Learning. Our study is the first step to connect these two classical mathematical models: Gaussian Elimination and Markov Decision Process. Our learning-based algorithm could help improve the performance of sparse solver, which has been verified in some numerical experiments.
Robust reinforcement learning (RL) is to find a policy that optimizes the worst-case performance over an uncertainty set of MDPs. In this paper, we focus on model-free robust RL, where the uncertainty set is defined to be centering at a misspecified MDP that generates a single sample trajectory sequentially and is assumed to be unknown. We develop a sample-based approach to estimate the unknown uncertainty set and design a robust Q-learning algorithm (tabular case) and robust TDC algorithm (function approximation setting), which can be implemented in an online and incremental fashion. For the robust Q-learning algorithm, we prove that it converges to the optimal robust Q function, and for the robust TDC algorithm, we prove that it converges asymptotically to some stationary points. Unlike the results in [Roy et al., 2017], our algorithms do not need any additional conditions on the discount factor to guarantee the convergence. We further characterize the finite-time error bounds of the two algorithms and show that both the robust Q-learning and robust TDC algorithms converge as fast as their vanilla counterparts(within a constant factor). Our numerical experiments further demonstrate the robustness of our algorithms. Our approach can be readily extended to robustify many other algorithms, e.g., TD, SARSA, and other GTD algorithms.
Intelligent traffic signal controllers, applying DQN algorithms to traffic light policy optimization, efficiently reduce traffic congestion by adjusting traffic signals to real-time traffic. Most propositions in the literature however consider that all vehicles at the intersection are detected, an unrealistic scenario. Recently, new wireless communication technologies have enabled cost-efficient detection of connected vehicles by infrastructures. With only a small fraction of the total fleet currently equipped, methods able to perform under low detection rates are desirable. In this paper, we propose a deep reinforcement Q-learning model to optimize traffic signal control at an isolated intersection, in a partially observable environment with connected vehicles. First, we present the novel DQN model within the RL framework. We introduce a new state representation for partially observable environments and a new reward function for traffic signal control, and provide a network architecture and tuned hyper-parameters. Second, we evaluate the performances of the model in numerical simulations on multiple scenarios, in two steps. At first in full detection against existing actuated controllers, then in partial detection with loss estimates for proportions of connected vehicles. Finally, from the obtained results, we define thresholds for detection rates with acceptable and optimal performance levels.
We use reinforcement learning to tackle the problem of untangling braids. We experiment with braids with 2 and 3 strands. Two competing players learn to tangle and untangle a braid. We interface the braid untangling problem with the OpenAI Gym environment, a widely used way of connecting agents to reinforcement learning problems. The results provide evidence that the more we train the system, the better the untangling player gets at untangling braids. At the same time, our tangling player produces good examples of tangled braids.
Double Q-learning is a classical method for reducing overestimation bias, which is caused by taking maximum estimated values in the Bellman operation. Its variants in the deep Q-learning paradigm have shown great promise in producing reliable value prediction and improving learning performance. However, as shown by prior work, double Q-learning is not fully unbiased and suffers from underestimation bias. In this paper, we show that such underestimation bias may lead to multiple non-optimal fixed points under an approximate Bellman operator. To address the concerns of converging to non-optimal stationary solutions, we propose a simple but effective approach as a partial fix for the underestimation bias in double Q-learning. This approach leverages an approximate dynamic programming to bound the target value. We extensively evaluate our proposed method in the Atari benchmark tasks and demonstrate its significant improvement over baseline algorithms.
Deep reinforcement learning (RL) algorithms can learn complex policies to optimize agent operation over time. RL algorithms have shown promising results in solving complicated problems in recent years. However, their application on real-world physical systems remains limited. Despite the advancements in RL algorithms, the industries often prefer traditional control strategies. Traditional methods are simple, computationally efficient and easy to adjust. In this paper, we first propose a new Q-learning algorithm for continuous action space, which can bridge the control and RL algorithms and bring us the best of both worlds. Our method can learn complex policies to achieve long-term goals and at the same time it can be easily adjusted to address short-term requirements without retraining. Next, we present an approximation of our algorithm which can be applied to address short-term requirements of any pre-trained RL algorithm. The case studies demonstrate that both our proposed method as well as its practical approximation can achieve short-term and long-term goals without complex reward functions.
A smart home energy management system (HEMS) can contribute towards reducing the energy costs of customers; however, HEMS suffers from uncertainty in both energy generation and consumption patterns. In this paper, we propose a sequence to sequence (Seq2Seq) learning-based supply and load prediction along with reinforcement learning-based HEMS control. We investigate how the prediction method affects the HEMS operation. First, we use Seq2Seq learning to predict photovoltaic (PV) power and home devices' load. We then apply Q-learning for offline optimization of HEMS based on the prediction results. Finally, we test the online performance of the trained Q-learning scheme with actual PV and load data. The Seq2Seq learning is compared with VARMA, SVR, and LSTM in both prediction and operation levels. The simulation results show that Seq2Seq performs better with a lower prediction error and online operation performance.
Approximation of the value functions in value-based deep reinforcement learning induces overestimation bias, resulting in suboptimal policies. We show that when the reinforcement signals received by the agents have a high variance, deep actor-critic approaches that overcome the overestimation bias lead to a substantial underestimation bias. We first address the detrimental issues in the existing approaches that aim to overcome such underestimation error. Then, through extensive statistical analysis, we introduce a novel, parameter-free Deep Q-learning variant to reduce this underestimation bias in deterministic policy gradients. By sampling the weights of a linear combination of two approximate critics from a highly shrunk estimation bias interval, our Q-value update rule is not affected by the variance of the rewards received by the agents throughout learning. We test the performance of the introduced improvement on a set of MuJoCo and Box2D continuous control tasks and demonstrate that it considerably outperforms the existing approaches and improves the state-of-the-art by a significant margin.
Cell switching has been identified as a major approach to significantly reduce the energy consumption of Heterogeneous Networks (HetNets). The main idea behind cell switching is to turn off idle or lightly loaded Base Stations (BSs) and to offload their traffic to neighbouring active cell(s). However, the impact of the offloaded traffic on the power consumption of the neighbouring cell(s) has not been studied sufficiently in the literature, thereby leading to the development of sub-optimal cell switching mechanisms. In this work, we first considered a Control/Data Separated Architecture (CDSA) with a macro cell serving as the Control Base Station (CBS) and multiple small cells as Data Base Stations (DBS). Then, a Q-learning assisted cell switching algorithm is developed in order to determine the small cells to switch off by considering the increase in power consumption of the macro cell due to offloaded traffic from the sleeping cells. The capacity of the macro cell is also taken into consideration to ensure that the Quality of Service (QoS) requirements of users are maintained. Simulation results show that the proposed cell switching algorithm can achieve up to 50% reduction in the total energy consumption of the considered HetNet scenario.
In value-based deep reinforcement learning methods, approximation of value functions induces overestimation bias and leads to suboptimal policies. We show that in deep actor-critic methods that aim to overcome the overestimation bias, if the reinforcement signals received by the agent have a high variance, a significant underestimation bias arises. To minimize the underestimation, we introduce a parameter-free, novel deep Q-learning variant. Our Q-value update rule combines the notions behind Clipped Double Q-learning and Maxmin Q-learning by computing the critic objective through the nested combination of maximum and minimum operators to bound the approximate value estimates. We evaluate our modification on the suite of several OpenAI Gym continuous control tasks, improving the state-of-the-art in every environment tested.
During the training of a reinforcement learning (RL) agent, the distribution of training data is non-stationary as the agent's behavior changes over time. Therefore, there is a risk that the agent is overspecialized to a particular distribution and its performance suffers in the larger picture. Ensemble RL can mitigate this issue by learning a robust policy. However, it suffers from heavy computational resource consumption due to the newly introduced value and policy functions. In this paper, to avoid the notorious resources consumption issue, we design a novel and simple ensemble deep RL framework that integrates multiple models into a single model. Specifically, we propose the \underline{M}inimalist \underline{E}nsemble \underline{P}olicy \underline{G}radient framework (MEPG), which introduces minimalist ensemble consistent Bellman update by utilizing a modified dropout operator. MEPG holds ensemble property by keeping the dropout consistency of both sides of the Bellman equation. Additionally, the dropout operator also increases MEPG's generalization capability. Moreover, we theoretically show that the policy evaluation phase in the MEPG maintains two synchronized deep Gaussian Processes. To verify the MEPG framework's ability to generalize, we perform experiments on the gym simulator, which presents that the MEPG framework outperforms or achieves a similar level of performance as the current state-of-the-art ensemble methods and model-free methods without increasing additional computational resource costs.
We study a stochastic variant of the vehicle routing problem arising in the context of domestic donor collection services. The problem we consider combines the following attributes. Customers requesting services are variable, in the sense that the customers are stochastic but are not restricted to a predefined set, as they may appear anywhere in a given service area. Furthermore, demand volumes are stochastic and observed upon visiting the customer. The objective is to maximize the expected served demands while meeting vehicle capacity and time restrictions. We call this problem the VRP with a highly Variable Customer basis and Stochastic Demands (VRP-VCSD). For this problem, we first propose a Markov Decision Process (MDP) formulation representing the classical centralized decision-making perspective where one decision-maker establishes the routes of all vehicles. While the resulting formulation turns out to be intractable, it provides us with the ground to develop a new MDP formulation, which we call partially decentralized. In this formulation, the action-space is decomposed by vehicle. However, the decentralization is incomplete as we enforce identical vehicle-specific policies while optimizing the collective reward. We propose several strategies to reduce the dimension of the state and action spaces associated with the partially decentralized formulation. These yield a considerably more tractable problem, which we solve via Reinforcement Learning. In particular, we develop a Q-learning algorithm called DecQN, featuring state-of-the-art acceleration techniques. We conduct a thorough computational analysis. Results show that DecQN considerably outperforms three benchmark policies. Moreover, we show that our approach can compete with specialized methods developed for the particular case of the VRP-VCSD, where customer locations and expected demands are known in advance.
Current GNN-oriented NAS methods focus on the search for different layer aggregate components with shallow and simple architectures, which are limited by the 'over-smooth' problem. To further explore the benefits from structural diversity and depth of GNN architectures, we propose a GNN generation pipeline with a novel two-stage search space, which aims at automatically generating high-performance while transferable deep GNN models in a block-wise manner. Meanwhile, to alleviate the 'over-smooth' problem, we incorporate multiple flexible residual connection in our search space and apply identity mapping in the basic GNN layers. For the search algorithm, we use deep-q-learning with epsilon-greedy exploration strategy and reward reshaping. Extensive experiments on real-world datasets show that our generated GNN models outperforms existing manually designed and NAS-based ones.
We propose using regularization for Multi-Agent Reinforcement Learning rather than learning explicit cooperative structures called {\em Multi-Agent Regularized Q-learning} (MARQ). Many MARL approaches leverage centralized structures in order to exploit global state information or removing communication constraints when the agents act in a decentralized manner. Instead of learning redundant structures which is removed during agent execution, we propose instead to leverage shared experiences of the agents to regularize the individual policies in order to promote structured exploration. We examine several different approaches to how MARQ can either explicitly or implicitly regularize our policies in a multi-agent setting. MARQ aims to address these limitations in the MARL context through applying regularization constraints which can correct bias in off-policy out-of-distribution agent experiences and promote diverse exploration. Our algorithm is evaluated on several benchmark multi-agent environments and we show that MARQ consistently outperforms several baselines and state-of-the-art algorithms; learning in fewer steps and converging to higher returns.
This paper introduces Greedy UnMix (GUM) for cooperative multi-agent reinforcement learning (MARL). Greedy UnMix aims to avoid scenarios where MARL methods fail due to overestimation of values as part of the large joint state-action space. It aims to address this through a conservative Q-learning approach through restricting the state-marginal in the dataset to avoid unobserved joint state action spaces, whilst concurrently attempting to unmix or simplify the problem space under the centralized training with decentralized execution paradigm. We demonstrate the adherence to Q-function lower bounds in the Q-learning for MARL scenarios, and demonstrate superior performance to existing Q-learning MARL approaches as well as more general MARL algorithms over a set of benchmark MARL tasks, despite its relative simplicity compared with state-of-the-art approaches.
Network slicing is a critical technique for 5G communications that covers radio access network (RAN), edge, transport and core slicing.The evolving network architecture requires the orchestration of multiple network resources such as radio and cache resources. In recent years, machine learning (ML) techniques have been widely applied for network management. However, most existing works do not take advantage of the knowledge transfer capability in ML. In this paper, we propose a deep transfer reinforcement learning (DTRL) scheme for joint radio and cache resource allocation to serve 5G RAN slicing. We first define a hierarchical architecture for joint resource allocation. Then we propose two DTRL algorithms: Q-value-based deep transfer reinforcement learning (QDTRL) and action selection-based deep transfer reinforcement learning (ADTRL). In the proposed schemes, learner agents utilize expert agents' knowledge to improve their performance on current tasks. The proposed algorithms are compared with both the model-free exploration bonus deep Q-learning (EB-DQN) and the model-based priority proportional fairness and time-to-live (PPF-TTL) algorithms. Compared with EB-DQN, our proposed DTRL-based method presents 21.4% lower delay for Ultra Reliable Low Latency Communications (URLLC) slice and 22.4% higher throughput for enhanced Mobile Broad Band (eMBB) slice, while achieving significantly faster convergence than EB-DQN. Moreover, 40.8% lower URLLC delay and 59.8% higher eMBB throughput are observed with respect to PPF-TTL.