We study reinforcement learning (RL) with linear function approximation. Existing algorithms for this problem only have high-probability regret and/or Probably Approximately Correct (PAC) sample complexity guarantees, which cannot guarantee the convergence to the optimal policy. In this paper, in order to overcome the limitation of existing algorithms, we propose a new algorithm called FLUTE, which enjoys uniform-PAC convergence to the optimal policy with high probability. The uniform-PAC guarantee is the strongest possible guarantee for reinforcement learning in the literature, which can directly imply both PAC and high probability regret bounds, making our algorithm superior to all existing algorithms with linear function approximation. At the core of our algorithm is a novel minimax value function estimator and a multi-level partition scheme to select the training samples from historical observations. Both of these techniques are new and of independent interest.

Differential equations in general and neural ODEs in particular are an essential technique in continuous-time system identification. While many deterministic learning algorithms have been designed based on numerical integration via the adjoint method, many downstream tasks such as active learning, exploration in reinforcement learning, robust control, or filtering require accurate estimates of predictive uncertainties. In this work, we propose a novel approach towards estimating epistemically uncertain neural ODEs, avoiding the numerical integration bottleneck. Instead of modeling uncertainty in the ODE parameters, we directly model uncertainties in the state space. Our algorithm - distributional gradient matching (DGM) - jointly trains a smoother and a dynamics model and matches their gradients via minimizing a Wasserstein loss. Our experiments show that, compared to traditional approximate inference methods based on numerical integration, our approach is faster to train, faster at predicting previously unseen trajectories, and in the context of neural ODEs, significantly more accurate.

Ensembles of deep neural networks have achieved great success recently, but they do not offer a proper Bayesian justification. Moreover, while they allow for averaging of predictions over several hypotheses, they do not provide any guarantees for their diversity, leading to redundant solutions in function space. In contrast, particle-based inference methods, such as Stein variational gradient descent (SVGD), offer a Bayesian framework, but rely on the choice of a kernel to measure the similarity between ensemble members. In this work, we study different SVGD methods operating in the weight space, function space, and in a hybrid setting. We compare the SVGD approaches to other ensembling-based methods in terms of their theoretical properties and assess their empirical performance on synthetic and real-world tasks. We find that SVGD using functional and hybrid kernels can overcome the limitations of deep ensembles. It improves on functional diversity and uncertainty estimation and approaches the true Bayesian posterior more closely. Moreover, we show that using stochastic SVGD updates, as opposed to the standard deterministic ones, can further improve the performance.

We develop nested variational inference (NVI), a family of methods that learn proposals for nested importance samplers by minimizing an forward or reverse KL divergence at each level of nesting. NVI is applicable to many commonly-used importance sampling strategies and provides a mechanism for learning intermediate densities, which can serve as heuristics to guide the sampler. Our experiments apply NVI to (a) sample from a multimodal distribution using a learned annealing path (b) learn heuristics that approximate the likelihood of future observations in a hidden Markov model and (c) to perform amortized inference in hierarchical deep generative models. We observe that optimizing nested objectives leads to improved sample quality in terms of log average weight and effective sample size.

Recent machine learning advances have proposed black-box estimation of unknown continuous-time system dynamics directly from data. However, earlier works are based on approximative ODE solutions or point estimates. We propose a novel Bayesian nonparametric model that uses Gaussian processes to infer posteriors of unknown ODE systems directly from data. We derive sparse variational inference with decoupled functional sampling to represent vector field posteriors. We also introduce a probabilistic shooting augmentation to enable efficient inference from arbitrarily long trajectories. The method demonstrates the benefit of computing vector field posteriors, with predictive uncertainty scores outperforming alternative methods on multiple ODE learning tasks.

Recent discoveries in Deep Neural Networks are allowing researchers to tackle some very complex problems such as image classification and audio classification, with improved theoretical and empirical justifications. This paper presents a novel scheme to incorporate the use of autoencoders in Fuzzy rule classifiers (FRC). Autoencoders when stacked can learn the complex non-linear relationships amongst data, and the proposed framework built towards FRC can allow users to input expert knowledge to the system. This paper further introduces four novel fine-tuning strategies for autoencoders to improve the FRC's classification and rule reduction performance. The proposed framework has been tested across five real-world benchmark datasets. Elaborate comparisons with over 15 previous studies, and across 10-fold cross validation performance, suggest that the proposed methods are capable of building FRCs which can provide state of the art accuracies.

The technological revolution taking place in the fields of robotics and artificial intelligence seems to indicate a future shift in our human-centred social paradigm towards a greater inclusion of artificial cognitive agents in our everyday environments. This means that collaborative scenarios between humans and robots will become more frequent and will have a deeper impact on everyday life. In this setting, research regarding trust in human–robot interactions (HRI) assumes a major importance in order to ensure the highest quality of the interaction itself, as trust directly affects the willingness of people to accept information produced by a robot and to cooperate with it. Many studies have already explored trust that humans give to robots and how this can be enhanced by tuning both the design and the behaviour of the machine, but not so much research has focused on the opposite scenario, that is the trust that artificial agents can assign to people. Despite this, the latter is a critical factor in joint tasks where humans and robots depend on each other's effort to achieve a shared goal: whereas a robot can fail, so can a person. For an artificial agent to know when to trust or distrust somebody and adapt its plans to this prediction can make all the difference in the success or failure of the task. Our work is centred on the design and development of an artificial cognitive architecture for a humanoid autonomous robot that incorporates trust, theory of mind (ToM) and episodic memory, as we believe these are the three key factors for the purpose of estimating the trustworthiness of others. We have tested our architecture on an established developmental psychology experiment [1] and the results we obtained confirm that our approach successfully models trust mechanisms and dynamics in cognitive robots. Trust is a fundamental, unavoidable component of social interactions that can be defined as the willingness of a party (the trustor) to rely on the actions of another party (the trustee), with the former having no control over the latter [2].

We present a new experiment demonstrating destructive interference in customers' estimates of conditional probabilities of product failure. We take the perspective of a manufacturer of consumer products, and consider two situations of cause and effect. Whereas individually the effect of the causes is similar, it is observed that when combined, the two causes produce the opposite effect. Such negative interference of two or more reasons may be exploited for better modeling the cognitive processes taking place in the customers' mind. Doing so can enhance the likelihood that a manufacturer will be able to design a better product, or a feature within it. Quantum probability has been used to explain some commonly observed deviations such as question order and response replicability effects, as well as in explaining paradoxes such as violations of the sure-thing principle, and Machina and Ellsberg paradoxes. In this work, we present results from a survey conducted regarding the effect of multiple observed symptoms on the drivability of a vehicle. We demonstrate that the set of responses cannot be explained using classical probability, but quantum formulation easily models it, as it allows for both positive and negative "interference" between events. Since quantum formulism also accounts for classical probability's predictions, it serves as a richer paradigm for modeling decision making behavior in engineering design and behavioral economics.

Learning generative probabilistic models is a core problem in machine learning, which presents significant challenges due to the curse of dimensionality. This paper proposes a joint dimensionality reduction and non-parametric density estimation framework, using a novel estimator that can explicitly capture the underlying distribution of appropriate reduced-dimension representations of the input data. The idea is to jointly design a nonlinear dimensionality reducing auto-encoder to model the training data in terms of a parsimonious set of latent random variables, and learn a canonical low-rank tensor model of the joint distribution of the latent variables in the Fourier domain. The proposed latent density model is non-parametric and universal, as opposed to the predefined prior that is assumed in variational auto-encoders. Joint optimization of the auto-encoder and the latent density estimator is pursued via a formulation which learns both by minimizing a combination of the negative log-likelihood in the latent domain and the auto-encoder reconstruction loss. We demonstrate that the proposed model achieves very promising results on toy, tabular, and image datasets on regression tasks, sampling, and anomaly detection.

We propose a fully Bayesian framework for learning ground truth labels from noisy annotators. Our framework ensures scalability by factoring a generative, Bayesian soft clustering model over label distributions into the classic David and Skene joint annotator-data model. Earlier research along these lines has neither fully incorporated label distributions nor explored clustering by annotators only or data only. Our framework incorporates all of these properties as: (1) a graphical model designed to provide better ground truth estimates of annotator responses as input to \emph{any} black box supervised learning algorithm, and (2) a standalone neural model whose internal structure captures many of the properties of the graphical model. We conduct supervised learning experiments using both models and compare them to the performance of one baseline and a state-of-the-art model.