Efficient exploration remains a major challenge for reinforcement learning. One reason is that the variability of the returns often depends on the current state and action, and is therefore heteroscedastic. Classical exploration strategies such as upper confidence bound algorithms and Thompson sampling fail to appropriately account for heteroscedasticity, even in the bandit setting. Motivated by recent findings that address this issue in bandits, we propose to use Information-Directed Sampling (IDS) for exploration in reinforcement learning. As our main contribution, we build on recent advances in distributional reinforcement learning and propose a novel, tractable approximation of IDS for deep Q-learning. The resulting exploration strategy explicitly accounts for both parametric uncertainty and heteroscedastic observation noise. We evaluate our method on Atari games and demonstrate a significant improvement over alternative approaches.
Generative Adversarial Networks (GANs) have shown great results in accurately modeling complex distributions, but their training is known to be difficult due to instabilities caused by a challenging minimax optimization problem. This is especially troublesome given the lack of an evaluation metric that can reliably detect non-convergent behaviors. We leverage the notion of duality gap from game theory in order to propose a novel convergence metric for GANs that has low computational cost. We verify the validity of the proposed metric for various test scenarios commonly used in the literature.
The energy output of photovoltaic (PV) power plants depends on the environment and thus fluctuates over time. As a result, PV power can cause instability in the power grid, in particular when increasingly used. Limiting the rate of change of the power output is a common way to mitigate these fluctuations, often with the help of large batteries. A reactive controller that uses these batteries to compensate ramps works in practice, but causes stress on the battery due to a high energy throughput. In this paper, we present a deep learning approach that uses images of the sky to compensate power fluctuations predictively and reduces battery stress. In particular, we show that the optimal control policy can be computed using information that is only available in hindsight. Based on this, we use imitation learning to train a neural network that approximates this hindsight-optimal policy, but uses only currently available sky images and sensor data. We evaluate our method on a large dataset of measurements and images from a real power plant and show that the trained policy reduces stress on the battery.
Learning-based methods have been successful in solving complex control tasks without significant prior knowledge about the system. However, these methods typically do not provide any safety guarantees, which prevents their use in safety-critical, real-world applications. In this paper, we present a learning-based model predictive control scheme that can provide provable high-probability safety guarantees. To this end, we exploit regularity assumptions on the dynamics in terms of a Gaussian process prior to construct provably accurate confidence intervals on predicted trajectories. Unlike previous approaches, we do not assume that model uncertainties are independent. Based on these predictions, we guarantee that trajectories satisfy safety constraints. Moreover, we use a terminal set constraint to recursively guarantee the existence of safe control actions at every iteration. In our experiments, we show that the resulting algorithm can be used to safely and efficiently explore and learn about dynamic systems.
We draw attention to an important, yet largely overlooked aspect of evaluating fairness for automated decision making systems---namely risk and welfare considerations. Our proposed family of measures corresponds to the long-established formulations of cardinal social welfare in economics, and is justified by the Rawlsian conception of fairness behind a veil of ignorance. The convex formulation of our welfare-based measures of fairness allows us to integrate them as a constraint into any convex loss minimization pipeline. Our empirical analysis reveals interesting trade-offs between our proposal and (a) prediction accuracy, (b) group discrimination, and (c) Dwork et al.'s notion of individual fairness. Furthermore and perhaps most importantly, our work provides both heuristic justification and empirical evidence suggesting that a lower-bound on our measures often leads to bounded inequality in algorithmic outcomes; hence presenting the first computationally feasible mechanism for bounding individual-level inequality.
How can we help a forgetful learner learn multiple concepts within a limited time frame? For long-term learning, it is crucial to devise teaching strategies that leverage the underlying forgetting mechanisms of the learner. In this paper, we cast the problem of adaptively teaching a forgetful learner as a novel discrete optimization problem, where we seek to optimize a natural objective function that characterizes the learner's expected performance throughout the teaching session. We then propose a simple greedy teaching strategy and derive strong performance guarantees based on two intuitive data-dependent properties, which capture the degree of diminishing returns of teaching each concept. We show that, given some assumptions about the learner's memory model, one can efficiently compute the performance bounds. Furthermore, we identify parameter settings of the memory model where the greedy strategy is guaranteed to achieve high performance. We demonstrate the effectiveness of our algorithm using extensive simulations along with user studies in two concrete applications, namely (i) an educational app for online vocabulary teaching and (ii) an app for teaching novices how to recognize animal species from images.
Learning algorithms have shown considerable prowess in simulation by allowing robots to adapt to uncertain environments and improve their performance. However, such algorithms are rarely used in practice on safety-critical systems, since the learned policy typically does not yield any safety guarantees. That is, the required exploration may cause physical harm to the robot or its environment. In this paper, we present a method to learn accurate safety certificates for nonlinear, closed-loop dynamical systems. Specifically, we construct a neural network Lyapunov function and a training algorithm that adapts it to the shape of the largest safe region in the state space. The algorithm relies only on knowledge of inputs and outputs of the dynamics, rather than on any specific model structure. We demonstrate our method by learning the safe region of attraction for a simulated inverted pendulum. Furthermore, we discuss how our method can be used in safe learning algorithms together with statistical models of dynamical systems.
Equality of opportunity (EOP) is an extensively studied conception of fairness in political philosophy. In this work, we map recently proposed notions of algorithmic fairness to economic models of EOP. We formally show that through our proposed mapping, many existing definition of algorithmic fairness, such as predictive value parity and equality of odds, can be interpreted as special cases of EOP. In this respect, our work serves as a unifying moral framework for understanding existing notions of algorithmic fairness. Most importantly, this framework allows us to explicitly spell out the moral assumptions underlying each notion of fairness, and also interpret recent fairness impossibility results in a new light. Last but not least and inspired by luck egalitarian models of EOP, we propose a new, more general family of measures for algorithmic fairness. We empirically show that employing a measure of algorithmic (un)fairness when its underlying moral assumptions are not satisfied, can have devastating consequences on the subjects' welfare.
This work presents a case study of a learning-based approach for target driven map-less navigation. The underlying navigation model is an end-to-end neural network which is trained using a combination of expert demonstrations, imitation learning (IL) and reinforcement learning (RL). While RL and IL suffer from a large sample complexity and the distribution mismatch problem, respectively, we show that leveraging prior expert demonstrations for pre-training can reduce the training time to reach at least the same level of performance compared to plain RL by a factor of 5. We present a thorough evaluation of different combinations of expert demonstrations, different RL algorithms and reward functions, both in simulation and on a real robotic platform. Our results show that the final model outperforms both standalone approaches in the amount of successful navigation tasks. In addition, the RL reward function can be significantly simplified when using pre-training, e.g. by using a sparse reward only. The learned navigation policy is able to generalize to unseen and real-world environments.