arXiv subjects

Doina Precup

Publications and source records attributed to Doina Precup.

At least 199 records · Page 11Linked to original sources

arXiv2019

Uncertainty Aware Learning from Demonstrations in Multiple Contexts using Bayesian Neural Networks

Diversity of environments is a key challenge that causes learned robotic controllers to fail due to the discrepancies between the training and evaluation conditions. Training from demonstrations in various conditions can mitigate---but not completely prevent---such failures. Learned controllers such as neural networks typically do not have a notion of uncertainty that allows to diagnose an offset between training and testing conditions, and potentially intervene. In this work, we propose to use Bayesian Neural Networks, which have such a notion of uncertainty. We show that uncertainty can be leveraged to consistently detect situations in high-dimensional simulated and real robotic domains in which the performance of the learned controller would be sub-par. Also, we show that such an uncertainty based solution allows making an informed decision about when to invoke a fallback strategy. One fallback strategy is to request more data. We empirically show that providing data only when requested results in increased data-efficiency.

cs.RO↗

arXiv2019

The Termination Critic

In this work, we consider the problem of autonomously discovering behavioral abstractions, or options, for reinforcement learning agents. We propose an algorithm that focuses on the termination condition, as opposed to -- as is common -- the policy. The termination condition is usually trained to optimize a control objective: an option ought to terminate if another has better value. We offer a different, information-theoretic perspective, and propose that terminations should focus instead on the compressibility of the option's encoding -- arguably a key reason for using abstractions. To achieve this algorithmically, we leverage the classical options framework, and learn the option transition model as a "critic" for the termination condition. Using this model, we derive gradients that optimize the desired criteria. We show that the resulting options are non-trivial, intuitively meaningful, and useful for learning and planning.

cs.AI↗

arXiv2019

Deep Reinforcement Learning that Matters

In recent years, significant progress has been made in solving challenging problems across various domains using deep reinforcement learning (RL). Reproducing existing work and accurately judging the improvements offered by novel methods is vital to sustaining this progress. Unfortunately, reproducing results for state-of-the-art deep RL methods is seldom straightforward. In particular, non-determinism in standard benchmark environments, combined with variance intrinsic to the methods, can make reported results tough to interpret. Without significance metrics and tighter standardization of experimental reporting, it is difficult to determine whether improvements over the prior state-of-the-art are meaningful. In this paper, we investigate challenges posed by reproducibility, proper experimental techniques, and reporting procedures. We illustrate the variability in reported metrics and results when comparing against common baselines and suggest guidelines to make future results in deep RL more reproducible. We aim to spur discussion about how to ensure continued progress in the field by minimizing wasted effort stemming from results that are non-reproducible and easily misinterpreted.

cs.LG↗

arXiv2018

Clustering-Oriented Representation Learning with Attractive-Repulsive Loss

The standard loss function used to train neural network classifiers, categorical cross-entropy (CCE), seeks to maximize accuracy on the training data; building useful representations is not a necessary byproduct of this objective. In this work, we propose clustering-oriented representation learning (COREL) as an alternative to CCE in the context of a generalized attractive-repulsive loss framework. COREL has the consequence of building latent representations that collectively exhibit the quality of natural clustering within the latent space of the final hidden layer, according to a predefined similarity function. Despite being simple to implement, COREL variants outperform or perform equivalently to CCE in a variety of scenarios, including image and news article classification using both feed-forward and convolutional neural networks. Analysis of the latent spaces created with different similarity functions facilitates insights on the different use cases COREL variants can satisfy, where the Cosine-COREL variant makes a consistently clusterable latent space, while Gaussian-COREL consistently obtains better classification accuracy than CCE.

cs.LG↗

arXiv2018

Environments for Lifelong Reinforcement Learning

To achieve general artificial intelligence, reinforcement learning (RL) agents should learn not only to optimize returns for one specific task but also to constantly build more complex skills and scaffold their knowledge about the world, without forgetting what has already been learned. In this paper, we discuss the desired characteristics of environments that can support the training and evaluation of lifelong reinforcement learning agents, review existing environments from this perspective, and propose recommendations for devising suitable environments in the future.

cs.AI↗

arXiv2018

Learning Safe Policies with Expert Guidance

We propose a framework for ensuring safe behavior of a reinforcement learning agent when the reward function may be difficult to specify. In order to do this, we rely on the existence of demonstrations from expert policies, and we provide a theoretical framework for the agent to optimize in the space of rewards consistent with its existing knowledge. We propose two methods to solve the resulting optimization: an exact ellipsoid-based method and a method in the spirit of the "follow-the-perturbed-leader" algorithm. Our experiments demonstrate the behavior of our algorithm in both discrete and continuous problems. The trained agent safely avoids states with potential negative effects while imitating the behavior of the expert in the other states.

cs.LG↗

arXiv2018

Combined Reinforcement Learning via Abstract Representations

In the quest for efficient and robust reinforcement learning methods, both model-free and model-based approaches offer advantages. In this paper we propose a new way of explicitly bridging both approaches via a shared low-dimensional learned encoding of the environment, meant to capture summarizing abstractions. We show that the modularity brought by this approach leads to good generalization while being computationally efficient, with planning happening in a smaller latent state space. In addition, this approach recovers a sufficient low-dimensional representation of the environment, which opens up new strategies for interpretable AI, exploration and transfer learning.

cs.LG↗

arXiv2018

The Barbados 2018 List of Open Issues in Continual Learning

We want to make progress toward artificial general intelligence, namely general-purpose agents that autonomously learn how to competently act in complex environments. The purpose of this report is to sketch a research outline, share some of the most important open issues we are facing, and stimulate further discussion in the community. The content is based on some of our discussions during a week-long workshop held in Barbados in February 2018.

cs.AI↗

arXiv2018

Convergent Tree Backup and Retrace with Function Approximation

Off-policy learning is key to scaling up reinforcement learning as it allows to learn about a target policy from the experience generated by a different behavior policy. Unfortunately, it has been challenging to combine off-policy learning with function approximation and multi-step bootstrapping in a way that leads to both stable and efficient algorithms. In this work, we show that the \textsc{Tree Backup} and \textsc{Retrace} algorithms are unstable with linear function approximation, both in theory and in practice with specific examples. Based on our analysis, we then derive stable and efficient gradient-based algorithms using a quadratic convex-concave saddle-point formulation. By exploiting the problem structure proper to these algorithms, we are able to provide convergence guarantees and finite-sample bounds. The applicability of our new analysis also goes beyond \textsc{Tree Backup} and \textsc{Retrace} and allows us to provide new convergence rates for the GTD and GTD2 algorithms without having recourse to projections or Polyak averaging.

cs.LG↗

arXiv2018

Exploring Uncertainty Measures in Deep Networks for Multiple Sclerosis Lesion Detection and Segmentation

Deep learning (DL) networks have recently been shown to outperform other segmentation methods on various public, medical-image challenge datasets [3,11,16], especially for large pathologies. However, in the context of diseases such as Multiple Sclerosis (MS), monitoring all the focal lesions visible on MRI sequences, even very small ones, is essential for disease staging, prognosis, and evaluating treatment efficacy. Moreover, producing deterministic outputs hinders DL adoption into clinical routines. Uncertainty estimates for the predictions would permit subsequent revision by clinicians. We present the first exploration of multiple uncertainty estimates based on Monte Carlo (MC) dropout [4] in the context of deep networks for lesion detection and segmentation in medical images. Specifically, we develop a 3D MS lesion segmentation CNN, augmented to provide four different voxel-based uncertainty measures based on MC dropout. We train the network on a proprietary, large-scale, multi-site, multi-scanner, clinical MS dataset, and compute lesion-wise uncertainties by accumulating evidence from voxel-wise uncertainties within detected lesions. We analyze the performance of voxel-based segmentation and lesion-level detection by choosing operating points based on the uncertainty. Empirical evidence suggests that uncertainty measures consistently allow us to choose superior operating points compared only using the network's sigmoid output as a probability.

cs.CV↗

arXiv2018

A Semi-Markov Chain Approach to Modeling Respiratory Patterns Prior to Extubation in Preterm Infants

After birth, extremely preterm infants often require specialized respiratory management in the form of invasive mechanical ventilation (IMV). Protracted IMV is associated with detrimental outcomes and morbidities. Premature extubation, on the other hand, would necessitate reintubation which is risky, technically challenging and could further lead to lung injury or disease. We present an approach to modeling respiratory patterns of infants who succeeded extubation and those who required reintubation which relies on Markov models. We compare the use of traditional Markov chains to semi-Markov models which emphasize cross-pattern transitions and timing information, and to multi-chain Markov models which can concisely represent non-stationarity in respiratory behavior over time. The models we developed expose specific, unique similarities as well as vital differences between the two populations.

eess.SP↗

arXiv2018

Predicting Extubation Readiness in Extreme Preterm Infants based on Patterns of Breathing

Extremely preterm infants commonly require intubation and invasive mechanical ventilation after birth. While the duration of mechanical ventilation should be minimized in order to avoid complications, extubation failure is associated with increases in morbidities and mortality. As part of a prospective observational study aimed at developing an accurate predictor of extubation readiness, Markov and semi-Markov chain models were applied to gain insight into the respiratory patterns of these infants, with more robust time-series modeling using semi-Markov models. This model revealed interesting similarities and differences between newborns who succeeded extubation and those who failed. The parameters of the model were further applied to predict extubation readiness via generative (joint likelihood) and discriminative (support vector machine) approaches. Results showed that up to 84\% of infants who failed extubation could have been accurately identified prior to extubation.

cs.LG↗

arXiv2018

Undersampling and Bagging of Decision Trees in the Analysis of Cardiorespiratory Behavior for the Prediction of Extubation Readiness in Extremely Preterm Infants

Extremely preterm infants often require endotracheal intubation and mechanical ventilation during the first days of life. Due to the detrimental effects of prolonged invasive mechanical ventilation (IMV), clinicians aim to extubate infants as soon as they deem them ready. Unfortunately, existing strategies for prediction of extubation readiness vary across clinicians and institutions, and lead to high reintubation rates. We present an approach using Random Forest classifiers for the analysis of cardiorespiratory variability to predict extubation readiness. We address the issue of data imbalance by employing random undersampling of examples from the majority class before training each Decision Tree in a bag. By incorporating clinical domain knowledge, we further demonstrate that our classifier could have identified 71% of infants who failed extubation, while maintaining a success detection rate of 78%.

cs.LG↗

arXiv2018

Attend Before you Act: Leveraging human visual attention for continual learning

When humans perform a task, such as playing a game, they selectively pay attention to certain parts of the visual input, gathering relevant information and sequentially combining it to build a representation from the sensory data. In this work, we explore leveraging where humans look in an image as an implicit indication of what is salient for decision making. We build on top of the UNREAL architecture in DeepMind Lab's 3D navigation maze environment. We train the agent both with original images and foveated images, which were generated by overlaying the original images with saliency maps generated using a real-time spectral residual technique. We investigate the effectiveness of this approach in transfer learning by measuring performance in the context of noise in the environment.

cs.AI↗

arXiv2018

Resolving Event Coreference with Supervised Representation Learning and Clustering-Oriented Regularization

We present an approach to event coreference resolution by developing a general framework for clustering that uses supervised representation learning. We propose a neural network architecture with novel Clustering-Oriented Regularization (CORE) terms in the objective function. These terms encourage the model to create embeddings of event mentions that are amenable to clustering. We then use agglomerative clustering on these embeddings to build event coreference chains. For both within- and cross-document coreference on the ECB+ corpus, our model obtains better results than models that require significantly more pre-annotated information. This work provides insight and motivating results for a new general approach to solving coreference and clustering problems with representation learning.

cs.CL↗

arXiv2018

Dyna Planning using a Feature Based Generative Model

Dyna-style reinforcement learning is a powerful approach for problems where not much real data is available. The main idea is to supplement real trajectories, or sequences of sampled states over time, with simulated ones sampled from a learned model of the environment. However, in large state spaces, the problem of learning a good generative model of the environment has been open so far. We propose to use deep belief networks to learn an environment model for use in Dyna. We present our approach and validate it empirically on problems where the state observations consist of images. Our results demonstrate that using deep belief networks, which are full generative models, significantly outperforms the use of linear expectation models, proposed in Sutton et al. (2008)

cs.LG↗

arXiv2018

Disentangling the independently controllable factors of variation by interacting with the world

It has been postulated that a good representation is one that disentangles the underlying explanatory factors of variation. However, it remains an open question what kind of training framework could potentially achieve that. Whereas most previous work focuses on the static setting (e.g., with images), we postulate that some of the causal factors could be discovered if the learner is allowed to interact with its environment. The agent can experiment with different actions and observe their effects. More specifically, we hypothesize that some of these factors correspond to aspects of the environment which are independently controllable, i.e., that there exists a policy and a learnable feature for each such aspect of the environment, such that this policy can yield changes in that feature with minimal changes to other features that explain the statistical variations in the observed data. We propose a specific objective function to find such factors, and verify experimentally that it can indeed disentangle independently controllable aspects of the environment without any extrinsic reward signal.

stat.ML↗

arXiv2018

Learning Robust Options

Robust reinforcement learning aims to produce policies that have strong guarantees even in the face of environments/transition models whose parameters have strong uncertainty. Existing work uses value-based methods and the usual primitive action setting. In this paper, we propose robust methods for learning temporally abstract actions, in the framework of options. We present a Robust Options Policy Iteration (ROPI) algorithm with convergence guarantees, which learns options that are robust to model uncertainty. We utilize ROPI to learn robust options with the Robust Options Deep Q Network (RO-DQN) that solves multiple tasks and mitigates model misspecification due to model uncertainty. We present experimental results which suggest that policy iteration with linear features may have an inherent form of robustness when using coarse feature representations. In addition, we present experimental results which demonstrate that robustness helps policy iteration implemented on top of deep neural networks to generalize over a much broader range of dynamics than non-robust policy iteration.

cs.AI↗