Developing a contemporary optimal transport (OT) solver requires navigating trade-offs among several critical requirements: GPU parallelization, scalability to high-dimensional problems, theoretical convergence guarantees, empirical performance in terms of precision versus runtime, and numerical stability in practice. With these challenges in mind, we introduce a specialized truncated Newton algorithm for entropic-regularized OT. In addition to proving that locally quadratic convergence is possible without assuming a Lipschitz Hessian, we provide strategies to maximally exploit the high rate of local convergence in practice. Our GPU-parallel algorithm exhibits exceptionally favorable runtime performance, achieving high precision orders of magnitude faster than many existing alternatives. This is evidenced by wall-clock time experiments on 24 problem sets (12 datasets $\times$ 2 cost functions). The scalability of the algorithm is showcased on an extremely large OT problem with $n \approx 10^6$, solved approximately under weak entropic regularization.

Building deep reinforcement learning (RL) agents that find a good policy with few samples has proven notoriously challenging. To achieve sample efficiency, recent work has explored updating neural networks with large numbers of gradient steps for every new sample. While such high update-to-data (UTD) ratios have shown strong empirical performance, they also introduce instability to the training process. Previous approaches need to rely on periodic neural network parameter resets to address this instability, but restarting the training process is infeasible in many real-world applications and requires tuning the resetting interval. In this paper, we focus on one of the core difficulties of stable training with limited samples: the inability of learned value functions to generalize to unobserved on-policy actions. We mitigate this issue directly by augmenting the off-policy RL training process with a small amount of data generated from a learned world model. Our method, Model-Augmented Data for TD Learning (MAD-TD) uses small amounts of generated data to stabilize high UTD training and achieve competitive performance on the most challenging tasks in the DeepMind control suite. Our experiments further highlight the importance of employing a good model to generate data, MAD-TD's ability to combat value overestimation, and its practical stability gains for continued learning.

We study the problem of distributional reinforcement learning using categorical parametrisations and a KL divergence loss. Previous work analyzing categorical distributional RL has done so using a Cramér distance-based loss, simplifying the analysis but creating a theory-practice gap. We introduce a preconditioned version of the algorithm, and prove that it is guaranteed to converge. We further derive the asymptotic variance of the categorical estimates under different learning rate regimes, and compare to that of classical reinforcement learning. We finally empirically validate our theoretical results and perform an empirical investigation into the relative strengths of using KL losses, and derive a number of actionable insights for practitioners.

Many-shot jailbreaking circumvents the safety alignment of LLMs by exploiting their ability to process long input sequences. To achieve this, the malicious target prompt is prefixed with hundreds of fabricated conversational exchanges between the user and the model. These exchanges are randomly sampled from a pool of unsafe question-answer pairs, making it appear as though the model has already complied with harmful instructions. In this paper, we present PANDAS: a hybrid technique that improves many-shot jailbreaking by modifying these fabricated dialogues with Positive Affirmations, Negative Demonstrations, and an optimized Adaptive Sampling method tailored to the target prompt's topic. We also introduce ManyHarm, a dataset of harmful question–answer pairs, and demonstrate through extensive experiments that PANDAS significantly outperforms baseline methods in long-context scenarios. Through attention analysis, we provide insights into how long-context vulnerabilities are exploited and show how PANDAS further improves upon many-shot jailbreaking.

The idea of value-aware model learning, that models should produce accurate value estimates, has gained prominence in model-based reinforcement learning. The MuZero loss, which penalizes a model’s value function prediction compared to the ground-truth value function, has been utilized in several prominent empirical works in the literature. However, theoretical investigation into its strengths and weaknesses is limited. In this paper, we analyze the family of value-aware model learning losses, which includes the popular MuZero loss. We show that these losses, as normally used, are uncalibrated surrogate losses, which means that they do not always recover the correct model and value function. Building on this insight, we propose corrections to solve this issue. Furthermore, we investigate the interplay between the loss calibration, latent model architectures, and auxiliary losses that are commonly employed when training MuZero-style agents. We show that while deterministic models can be sufficient to predict accurate values, learning calibrated stochastic models is still advantageous.

Sampling multiple outputs from a Large Language Model (LLM) and selecting the most frequent (Self-consistency) or highest-scoring (Best-of-N) candidate is a popular approach to achieve higher accuracy in tasks with discrete final answers. Best-of-N (BoN) selects the output with the highest reward, and with perfect rewards, it often achieves near-perfect accuracy. With imperfect rewards from reward models, however, BoN fails to reliably find the correct answer and its performance degrades drastically. We consider the distribution of BoN’s outputs and highlight that, although the correct answer does not usually have a probability close to one under imperfect rewards, it is often the most likely outcome. This suggests that the mode of this distribution can be more reliably correct than a sample from it. Based on this idea, we propose Majority-of-the-Bests (MoB), a novel selection mechanism that estimates the output distribution of BoN via bootstrapping and selects its mode. Experimental results across five benchmarks, three different base LLMs, and two reward models demonstrate consistent improvements over BoN in 25 out of 30 setups. We also provide theoretical results for the consistency of the bootstrapping. MoB serves as a simple, yet strong alternative to BoN and self-consistency, and more broadly, motivates further research in more nuanced selection mechanisms.

We propose and theoretically analyze an approach for planning with an approximate model in reinforcement learning that can reduce the adverse impact of model error. If the model is accurate enough, it accelerates the convergence to the true value function too. One of its key components is the MaxEnt Model Correction (MoCo) procedure that corrects the model’s next-state distributions based on a Maximum Entropy density estimation formulation. Based on MoCo, we introduce the Model Correcting Value Iteration (MoCoVI) algorithm, and its sampled-based variant MoCoDyna. We show that MoCoVI and MoCoDyna’s convergence can be much faster than the conventional model-free algorithms. Unlike traditional model-based algorithms, MoCoVI and MoCoDyna effectively utilize an approximate model and still converge to the correct value function.

We consider the problem of learning models for risk-sensitive reinforcement learning. We theoretically demonstrate that proper value equivalence, a method of learning models which can be used to plan optimally in the risk-neutral setting, is not sufficient to plan optimally in the risk-sensitive setting. We leverage distributional reinforcement learning to introduce two new notions of model equivalence, one which is general and can be used to plan for any risk measure, but is intractable; and a practical variation which allows one to choose which risk measures they may plan optimally for. We demonstrate how our models can be used to augment any model-free risk-sensitive algorithm, and provide both tabular and large-scale experiments to demonstrate our method’s ability.

We introduce new planning and reinforcement learning algorithms for discounted MDPs that utilize an approximate model of the environment to accelerate the convergence of the value function. Inspired by the splitting approach in numerical linear algebra, we introduce \emph{Operator Splitting Value Iteration} (OS-VI) for both Policy Evaluation and Control problems. OS-VI achieves a much faster convergence rate when the model is accurate enough. We also introduce a sample-based version of the algorithm called OS-Dyna. Unlike the traditional Dyna architecture, OS-Dyna still converges to the correct value function in presence of model approximation error.

Prioritized Experience Replay (ER) has been empirically shown to improve sample efficiency across many domains and attracted great attention; however, there is little theoretical understanding of why such prioritized sampling helps and its limitations. In this work, we take a deep look at the prioritized ER. In a supervised learning setting, we show the equivalence between the error-based prioritized sampling method for minimizing mean squared error and the uniform sampling for cubic power loss. We then provide theoretical insight into why error-based prioritized sampling improves convergence rate upon uniform sampling when minimizing mean squared error during early learning. Based on the insight, we further point out two limitations of the prioritized ER method: 1) outdated priorities and 2) insufficient coverage of the sample space. To mitigate the limitations, we propose our model-based stochastic gradient Langevin dynamics sampling method. We show that our method does provide states distributed close to an ideal prioritized sampling distribution estimated by the brute-force method, which does not suffer from the two limitations. We conduct experiments on both discrete and continuous control problems to show our approach’s efficacy and examine the practical implication of our method in an autonomous driving application.

The convergence rate of Value Iteration (VI), a fundamental procedure in dynamic programming and reinforcement learning, for solving MDPs can be slow when the discount factor is close to one. We propose modifications to VI in order to potentially accelerate its convergence behaviour. The key insight is the realization that the evolution of the value function approximations $(V_k)_{k \geq 0}$ in the VI procedure can be seen as a dynamical system. This opens up the possibility of using techniques from \emph{control theory} to modify, and potentially accelerate, this dynamics. We present such modifications based on simple controllers, such as PD (Proportional-Derivative), PI (Proportional-Integral), and PID. We present the error dynamics of these variants of VI, and provably (for certain classes of MDPs) and empirically (for more general classes) show that the convergence rate can be significantly improved. We also propose a gain adaptation mechanism in order to automatically select the controller gains, and empirically show the effectiveness of this procedure.

For multi-valued functions---such as when the conditional distribution on targets given the inputs is multi-modal---standard regression approaches are not always desirable because they provide the conditional mean. Modal regression algorithms address this issue by instead finding the conditional mode(s). Most, however, are nonparametric approaches and so can be difficult to scale. Further, parametric approximators, like neural networks, facilitate learning complex relationships between inputs and targets. In this work, we propose a parametric modal regression algorithm. We use the implicit function theorem to develop an objective, for learning a joint function over inputs and targets. We empirically demonstrate on several synthetic problems that our method (i) can learn multi-valued functions and produce the conditional modes, (ii) scales well to high-dimensional inputs, and (iii) can even be more effective for certain uni-modal problems, particularly for high-frequency functions. We demonstrate that our method is competitive in a real-world modal regression problem and two regular regression datasets.

Dyna is an architecture for model based reinforcement learning (RL), where simulated experience from a model is used to update policies or value functions. A key component of Dyna is search control, the mechanism to generate the state and action from which the agent queries the model, which remains largely unexplored. In this work, we propose to generate such states by using the trajectory obtained from Hill Climbing (HC) the current estimate of the value function. This has the effect of propagating value from high value regions and of preemptively updating value estimates of the regions that the agent is likely to visit next. We derive a noisy projected natural gradient algorithm for hill climbing, and highlight a connection to Langevin dynamics. We provide an empirical demonstration on four classical domains that our algorithm, HC Dyna, can obtain significant sample efficiency improvements. We study the properties of different sampling distributions for search control, and find that there appears to be a benefit specifically from using the samples generated by climbing on current value estimates from low value to high value region.

This paper considers the problem of estimating the distribution of returns in reinforcement learning (i.e., distributional RL problem). It presents a new representational framework to maintain the uncertainty of returns and provides mathematical tools to compute it. We show that instead of representing a probability distribution function of returns, one can represent their characteristic function instead, the Fourier transform of their distribution. We call the new representation Characteristic Value Function (CVF), which can be interpreted as the frequency domain representation of the probability distribution of returns. We show that the CVF satisfies a Bellman-like equation, and its corresponding Bellman operator is contraction with respect to certain metrics. The contraction property allows us to devise an iterative procedure to compute the CVF, which we call Characteristic Value Iteration (CVI). We analyze CVI and its approximate variant and show how approximation errors affect the quality of computed CVF.

Recent work has shown that reinforcement learning (RL) is a promising approach to control dynamical systems described by partial differential equations (PDE). This paper shows how to use RL to tackle more general PDE control problems that have continuous high-dimensional action spaces with spatial relationship among action dimensions. In particular, we propose the concept of action descriptors, which encode regularities among spatially-extended action dimensions and enable the agent to control high-dimensional action PDEs. We provide theoretical evidence suggesting that this approach can be more sample efficient compared to a conventional approach that treats each action dimension separately and does not explicitly exploit the spatial regularity of the action space. The action descriptor approach is then used within the deep deterministic policy gradient algorithm. Experiments on two PDE control problems, with up to 256-dimensional continuous actions, show the advantage of the proposed approach over the conventional one.

This paper introduces a model-based reinforcement learning (MBRL) framework that incorporates the underlying decision problem in learning the transition model of the environment. This is in contrast with conventional approaches to MBRL that learn the model of the environment, for example by finding the maximum likelihood estimate, without taking into account the decision problem. Value-Aware Model Learning (VAML) framework argues that this might not be a good idea, especially if the true model of the environment does not belong to the model class from which we are estimating the model. The original VAML framework, however, may result in an optimization problem that is difficult to solve. This paper introduces a new MBRL class of algorithms, called Iterative VAML, that benefits from the structure of how the planning is performed (i.e., through approximate value iteration) to devise a simpler optimization problem. The paper theoretically analyzes Iterative VAML and provides finite sample error upper bound guarantee for it.

We consider the problem of estimating the transition probability kernel to be used by a model-based reinforcement learning (RL) algorithm. We argue that estimating a generative model that minimizes a probabilistic loss, such as the log-loss, is an overkill because it does not take into account the underlying structure of decision problem and the RL algorithm that intends to solve it. We introduce a loss function that takes the structure of the value function into account. We provide a finite-sample upper bound for the loss function showing the dependence of the error on model approximation error, number of samples, and the complexity of the model space. We also empirically compare the method with the maximum likelihood estimator on a simple problem.

We propose Random Projection Filter Bank (RPFB) as a generic and simple approach to extract features from time series data. RPFB is a set of randomly generated stable autoregressive filters that are convolved with the input time series to generate the features. These features can be used by any conventional machine learning algorithm for solving tasks such as time series prediction, classification with time series data, etc. Different filters in RPFB extract different aspects of the time series, and together they provide a reasonably good summary of the time series. RPFB is easy to implement, fast to compute, and parallelizable. We provide an error upper bound indicating that RPFB provides a reasonable approximation to a class of dynamical systems. The empirical results in a series of synthetic and real-world problems show that RPFB is an effective method to extract features from time series.

We propose a new class of computationally fast algorithms to find close to optimal policy for Markov Decision Processes (MDP) with large finite horizon T.The main idea is that instead of planning until the time horizon T, we plan only up to a truncated horizon H << T and use an estimate of the true optimal value function as the terminal value. Our approach of finding the terminal value function is to learn a mapping from an MDP to its value function by solving many similar MDPs during a training phase and fit a regression estimator. We analyze the method by providing an error propagation theorem that shows the effect of various sources of errors to the quality of the solution. We also empirically validate this approach in a real-world application of designing an energy management system for Hybrid Electric Vehicles with promising results.

We study two regularization-based approximate policy iteration algorithms, namely REG-LSPI and REG-BRM, to solve reinforcement learning and planning problems in discounted Markov Decision Processes with large state and finite action spaces. The core of these algorithms are the regularized extensions of the Least- Squares Temporal Difference (LSTD) learning and Bellman Residual Minimization (BRM), which are used in the algorithms' policy evaluation steps. Regularization provides a convenient way to control the complexity of the function space to which the estimated value function belongs and as a result enables us to work with rich nonparametric function spaces. We derive efficient implementations of our methods when the function space is a reproducing kernel Hilbert space. We analyze the statistical properties of REG-LSPI and provide an upper bound on the policy evaluation error and the performance loss of the policy returned by this method. Our bound shows the dependence of the loss on the number of samples, the capacity of the function space, and some intrinsic properties of the underlying Markov Decision Process. The dependence of the policy evaluation bound on the number of samples is minimax optimal. This is the first work that provides such a strong guarantee for a nonparametric approximate policy iteration algorithm. (This work is an extension of the NIPS 2008 conference paper by Farahmand et al. (2009b).)

Maximum entropy inverse optimal control (MaxEnt IOC) is an effective means of discovering the underlying cost function of demonstrated human activity and can be used to predict human behavior over low-dimensional state spaces (i.e., forecasting of 2D trajectories). To enable inference in very large state spaces, we introduce an approximate MaxEnt IOC procedure to address the fundamental computational bottleneck stemming from calculating the partition function via dynamic programming. Approximate MaxEnt IOC is based on two components: approximate dynamic programming and Monte Carlo sampling. We analyze this approximation approach and provide a finite-sample error upper bound on its excess loss. We validate the proposed method in the context of analyzing dual-agent interactions from video, where we use approximate MaxEnt IOC to simulate mental images of a single agents body pose sequence (a high-dimensional image space). We experiment with sequences image data taken from RGB and RGBD data and show that it is possible to learn cost functions that lead to accurate predictions in high-dimensional problems that were previously intractable.

We introduce a method for regularizing linearly parameterized functions using general derivative-based penalties, which relies on sampling as well as finite-difference approximations of the relevant derivatives. We call this approach sample-based approximate regularization (SAR). We provide theoretical guarantees on the fidelity of such regularizers, compared to those they approximate, and prove that the approximations converge efficiently. We also examine the empirical performance of SAR on several datasets.

This paper addresses the problem of automatic generation of features for value function approximation in reinforcement learning. Bellman Error Basis Functions (BEBFs) have been shown to improve the error of policy evaluation with function approximation, with a convergence rate similar to that of value iteration. We propose a simple, fast and robust algorithm based on random projections, which generates BEBFs for sparse feature spaces. We provide a finite sample analysis of the proposed method, and prove that projections logarithmic in the dimension of the original space guarantee a contraction in the error. Empirical results demonstrate the strength of this method in domains in which choosing a good state representation is challenging.

We propose an approach to learning from demonstration (LfD) which leverages expert data, even if the expert examples are very few or inaccurate. We achieve this by integrating LfD in an approximate policy iteration algorithm. The key idea of our approach is that expert examples are used to generate linear constraints on the optimization, in a similar fashion to large-margin classification. We prove an upper bound on the true Bellman error of the approximation computed by the algorithm at each iteration. We show empirically that the algorithm outperforms both pure policy iteration, as well as DAgger (a state-of-art LfD algorithm) and supervised learning in a variety of scenarios, including when very few and/or imperfect demonstrations are available. Our experiments include simulations as well as a real robotic navigation task.

Value Pursuit Iteration (VPI) is an approximate value iteration algorithm that finds a close to optimal policy for reinforcement learning and planning problems with large state spaces. VPI has two main features: First, it is a nonparametric algorithm that finds a good sparse approximation of the optimal value function given a dictionary of features. The algorithm is almost insensitive to the number of irrelevant features. Second, after each iteration of VPI, the algorithm adds a set of functions based on the currently learned value function to the dictionary. This increases the representation power of the dictionary in a way that is directly relevant to the goal of having a good approximation of the optimal value function. We theoretically study VPI and provide a finite-sample error upper bound for it.

Many practitioners of reinforcement learning problems have observed that oftentimes the performance of the agent reaches very close to the optimal performance even though the estimated (action-)value function is still far from the optimal one. The goal of this paper is to explain and formalize this phenomenon by introducing the concept of the action-gap regularity. As a typical result, we prove that for an agent following the greedy policy (\hat{\pi}) with respect to an action-value function (\hat{Q}), the performance loss (E[V^(X) - V^{\hat{X}} (X)]) is upper bounded by (O(|| \hat{Q} - Q^||_\infty^{1+\zeta})), in which (\zeta >= 0) is the parameter quantifying the action-gap regularity. For (\zeta > 0), our results indicate smaller performance loss compared to what previous analyses had suggested. Finally, we show how this regularity affects the performance of the family of approximate value iteration algorithms.

This paper addresses robust estimation of the uncalibrated visual-motor Jacobian for an image-based visual servoing (IBVS) system. The proposed method does not require knowledge of model or system parameters and is robust to outliers caused by various visual tracking errors, such as occlusion or mis-tracking. Previous uncalibrated methods are not robust to outliers and assume that the visual-motor data belong to the underlying model. In unstructured environments, this assumption may not hold. Outliers to the visual-motor model may deteriorate the Jacobian, which can make the system unstable or drive the arm in the wrong direction. We propose to apply a statistically robust M-estimator to reject the outliers. We compare the quality of the robust Jacobian estimation with the least squares-based estimation. The effect of outliers on the estimation quality is studied through MATLAB simulations and eye-in-hand visual servoing experiments using a WAM arm. Experimental results show that the Jacobian estimated by robust M-estimation is robust when up to 40% of the visual-motor data are outliers.

We address the question of how the approximation error/Bellman residual at each iteration of the Approximate Policy/Value Iteration algorithms influences the quality of the resulted policy. We quantify the performance loss as the Lp norm of the approximation error/Bellman residual at each iteration. Moreover, we show that the performance loss depends on the expectation of the squared Radon-Nikodym derivative of a certain distribution rather than its supremum -- as opposed to what has been suggested by the previous results. Also our results indicate that the contribution of the approximation/Bellman error to the performance loss is more prominent in the later iterations of API/AVI, and the effect of an error term in the earlier iterations decays exponentially fast.

Designing an intelligent situated agent is a difficult task because the designer must see the problem from the agent's viewpoint, considering all its sensors, actuators, and computation systems. In this paper, we introduce a bio-inspired hybridization of reinforcement learning, cooperative co-evolution, and a cultural-inspired memetic algorithm for the automatic development of behavior-based agents. Reinforcement learning is responsible for the individual-level adaptation. Cooperative co-evolution performs at the population level and provides basic decision-making modules for the reinforcement-learning procedure. The culture-based memetic algorithm, which is a new computational interpretation of the meme metaphor, increases the lifetime performance of agents by sharing learning experiences between all agents in the society. In this paper, the design problem is decomposed into two different parts: 1) developing a repertoire of behavior modules and 2) organizing them in the agent's architecture. Our proposed cooperative co-evolutionary approach solves the first problem by evolving behavior modules in their separate genetic pools. We address the problem of relating the fitness of the agent to the fitness of behavior modules by proposing two fitness sharing mechanisms, namelyuniformandvalue-basedfitness sharing mechanisms. The organization of behavior modules in the architecture is determined by our structure learning method. A mathematical formulation is provided that shows how to decompose the value of the structure into simpler components. These values are estimated during learning and are used to find the organization of behavior modules during the agent's lifetime. To accelerate the learning process, we introduce a culture-based method based on our new interpretation of the meme metaphor. Our proposed memetic algorithm is a mechanism for sharing learned structures among agents in the society. Lifetime performance of the agent, which is quite important for real-world applications, increases considerably when the memetic algorithm is in action. Finally, we apply our methods to two benchmark problems: an abstract problem and a decentralized multirobot object-lifting task, and we achieve human-competitive architecture designs.

To address the difficulty of designing a controller for complex visual-servoing tasks, two learning-based uncalibrated approaches are introduced. The first method starts by building an estimated model for the visual-motor forward kinematic of the vision-robot system by a locally linear regression method. Afterwards, it uses a reinforcement learning method named Regularized Fitted Q-Iteration to find a controller (i.e. policy) for the system (model-based RL). The second method directly uses samples coming from the robot without building any intermediate model (model-free RL). The simulation results show that both methods perform comparably well despite not having any a priori knowledge about the robot.

In this paper we consider approximate policy-iteration-based reinforcement learning algorithms. In order to implement a flexible function approximation scheme we propose the use of non-parametric methods with regularization, providing a convenient way to control the complexity of the function approximator. We propose two novel regularized policy iteration algorithms by adding L2-regularization to two widely-used policy evaluation methods: Bellman residual minimization (BRM) and least-squares temporal difference learning (LSTD). We derive efficient implementation for our algorithms when the approximate value-functions belong to a reproducing kernel Hilbert space. We also provide finite-sample performance bounds for our algorithms and show that they are able to achieve optimal rates of convergence under the studied conditions.

Intuitively, learning should be easier when the data points lie on a low-dimensional submanifold of the input space. Recently there has been a growing interest in algorithms that aim to exploit such geometrical properties of the data. Oftentimes these algorithms require estimating the dimension of the manifold first. In this paper we propose an algorithm for dimension estimation and study its finite-sample behaviour. The algorithm estimates the dimension locally around the data points using nearest neighbor techniques and then combines these local estimates. We show that the rate of convergence of the resulting estimate is independent of the dimension of the input space and hence the algorithm is “manifold-adaptive”. Thus, when the manifold supporting the data is low dimensional, the algorithm can be exponentially more efficient than its counterparts that are not exploiting this property. Our computer experiments confirm the obtained theoretical results. 1.

In this paper, we present two methods for the estimation of a globally valid visual-motor model of a robotic manipulator. In conventional uncalibrated visual servoing, the visuo-motor function is approximated locally with a Jacobian. However, for optimal task planning, or nonlinear controller design with global stability guarantee, one needs to know a model that provides some information about the behavior of the system over the whole workspace. Our presented methods remedy this drawback in uncalibrated visual servoing by incrementally building a global estimator based on the movement history. We implement two such methods. The first method is a K-nearest neighborhood regressor over Jacobian that uses previously estimated local models. The second method stores previous movements and computes an estimate of the Jacobian by solving a local least squares problem. Experimental results show that both methods provide better global estimation quality compared to the conventional local estimation method with much lower estimation variance.

Designing an intelligent situated agent is a difficult task as the designer must see the problem from the agent's standpoint considering all its sensors and actuators. We have devised a co-evolutionary/reinforcement learning hybrid method to automate the design of hierarchical behavior-based systems. In our approach, the design problem is decomposed into two separate parts: developing a repertoire of behaviors and organizing those behaviors in a structure. Mathematical formulation shows how to decompose the value of the structure to simpler components. These components can be estimated and used to find the optimal organization of behaviors during the agent's lifetime. Moreover, a novel co-evolutionary mechanism is suggested that evolves each type of behavior separately in their own genetic pool. Our method is applied to the decentralized multi-robot object lifting task which results in human-competitive performance.

Behavior-based systems have been successfully used in control and robotics applications. In traditional behavior-based systems, only a single behavior controls the agent in any time step. However, this behavior arbitration is not appropriate for many complex tasks. In this paper, we propose Hierarchical Soft Behavior-based Architecture that uses the concept of soft suppression to coordinate flexibly between behaviors. In our method, we use reinforcement learning to find an appropriate amount of suppression for each behavior in the architecture, in addition to learn the internal mechanism of each behavior. Several experiments are provided to show the effectiveness of our method in the mobile robot navigation task.

Channel assignment problem in cellular communication is a difficult combinatorial optimization problem. There is no exact polynomial-time solution for it and searching the whole solution space is infeasible for large problems. By defining the problem's cost function as the energy function of a chaotic Hopfield neural network, we devise a framework for finding competitive suboptimal or even optimal solutions for combinatorial optimization problem in general, and channel assignment problem in particular. In our architecture, we inject chaotic noise in order to help the network escape from local minima of the energy function while we enforce problem constraints by external inputs of neurons. Experimental results show the superiority of our method to other methods.

Hand-design of an intelligent agent's behaviors and their hierarchy is a very hard task. One of the most important steps toward creating intelligent agents is providing them with capability to learn the required behaviors and their architecture. Architecture learning in a behavior-based agent with subsumption architecture is considered in this paper. Overall value function is decomposed into easily calculate-able parts in order to learn the behavior hierarchy. Using probabilistic formulations, two different decomposition methods are discussed: storing the estimated value of each behavior in each layer, and storing the ordering of behaviors in the architecture. Using defined decompositions, two appropriate credit assignment methods are designed. Finally, the proposed methods are tested in a multi-robot object-lifting task that results in satisfactory performance.