Recursive stochastic algorithms have gained significant attention in the recent past due to data driven applications. Examples include stochastic gradient descent for solving large-scale optimization problems and empirical dynamic programming algorithms for solving Markov decision problems. These recursive stochastic algorithms approximates certain contraction operators and can be viewed within the framework of iterated random maps. Accordingly, we consider iterated random maps over a Polish space that simulates a contraction operator over that Polish space. Assume that the iterated maps are indexed by $n$ such that as $n\rightarrow\infty$, each realization of the random map converges (in some sense) to the contraction map it is simulating. We show that starting from the same initial condition, the distribution of the random sequence generated by the iterated random maps converge weakly to the trajectory generated by the contraction operator. We further show that under certain conditions, the time average of the random sequence converge to the spatial mean of the invariant distribution. We then apply these results to logistic regression, empirical value iteration, empirical Q value iteration, and empirical relative value iteration for finite state finite action MDPs.

In an episodic Markov Decision Process (MDP) problem, an online algorithm chooses from a set of actions in a sequence of $H$ trials, where $H$ is the episode length, in order to maximize the total payoff of the chosen actions. Q-learning, as the most popular model-free reinforcement learning (RL) algorithm, directly parameterizes and updates value functions without explicitly modeling the environment. Recently, [Jin et al. 2018] studies the sample complexity of Q-learning with finite states and actions. Their algorithm achieves nearly optimal regret, which shows that Q-learning can be made sample efficient. However, MDPs with large discrete states and actions [Silver et al. 2016] or continuous spaces [Mnih et al. 2013] cannot learn efficiently in this way. Hence, it is critical to develop new algorithms to solve this dilemma with provable guarantee on the sample complexity. With this motivation, we propose a novel algorithm that works for MDPs with a more general setting, which has infinitely many states and actions and assumes that the payoff function and transition kernel are Lipschitz continuous. We also provide corresponding theory justification for our algorithm. It achieves the regret $\tilde{\mathcal{O}}(K^{\frac{d+1}{d+2}}\sqrt{H^3}),$ where $K$ denotes the number of episodes and $d$ denotes the dimension of the joint space. To the best of our knowledge, this is the first analysis in the model-free setting whose established regret matches the lower bound up to a logarithmic factor.

Multimodal learning has been lacking principled ways of combining information from different modalities and learning a low-dimensional manifold of meaningful representations. We study multimodal learning and sensor fusion from a latent variable perspective. We first present a regularized recurrent attention filter for sensor fusion. This algorithm can dynamically combine information from different types of sensors in a sequential decision making task. Each sensor is bonded with a modular neural network to maximize utility of its own information. A gating modular neural network dynamically generates a set of mixing weights for outputs from sensor networks by balancing utility of all sensors' information. We design a co-learning mechanism to encourage co-adaption and independent learning of each sensor at the same time, and propose a regularization based co-learning method. In the second part, we focus on recovering the manifold of latent representation. We propose a co-learning approach using probabilistic graphical model which imposes a structural prior on the generative model: multimodal variational RNN (MVRNN) model, and derive a variational lower bound for its objective functions. In the third part, we extend the siamese structure to sensor fusion for robust acoustic event detection. We perform experiments to investigate the latent representations that are extracted; works will be done in the following months. Our experiments show that the recurrent attention filter can dynamically combine different sensor inputs according to the information carried in the inputs. We consider MVRNN can identify latent representations that are useful for many downstream tasks such as speech synthesis, activity recognition, and control and planning. Both algorithms are general frameworks which can be applied to other tasks where different types of sensors are jointly used for decision making.

This work tackles the problem of robust zero-shot planning in non-stationary stochastic environments. We study Markov Decision Processes (MDPs) evolving over time and consider Model-Based Reinforcement Learning algorithms in this setting. We make two hypotheses: 1) the environment evolves continuously and its evolution rate is bounded, 2) a current model is known at each decision epoch but not its evolution. Our contribution can be presented in four points. First, we define this specific class of MDPs that we call Non-Stationary MDPs (NSMDPs). We introduce the notion of regular evolution by making an hypothesis of Lipschitz-Continuity on the transition and reward functions w.r.t. time. Secondly, we consider a planning agent using the current model of the environment, but unaware of its future evolution. This leads us to consider a worst-case method where the environment is seen as an adversarial agent. Third, following this approach, we propose the Risk-Averse Tree-Search (RATS) algorithm. This is a zero-shot Model-Based method similar to Minimax search. Finally, we illustrate the benefits brought by RATS empirically and compare its performance with reference Model-Based algorithms.

Decomposition methods have been proposed to approximate solutions to large sequential decision making problems. In contexts where an agent interacts with multiple entities, utility decomposition can be used to separate the global objective into local tasks considering each individual entity independently. An arbitrator is then responsible for combining the individual utilities and selecting an action in real time to solve the global problem. Although these techniques can perform well empirically, they rely on strong assumptions of independence between the local tasks and sacrifice the optimality of the global solution. This paper proposes an approach that improves upon such approximate solutions by learning a correction term represented by a neural network. We demonstrate this approach on a fisheries management problem where multiple boats must coordinate to maximize their catch over time as well as on a pedestrian avoidance problem for autonomous driving. In each problem, decomposition methods can scale to multiple boats or pedestrians by using strategies involving one entity. We verify empirically that the proposed correction method significantly improves the decomposition method and outperforms a policy trained on the full scale problem without utility decomposition.

Decision trees are flexible models that are well suited for many statistical regression problems. In a Bayesian framework for regression trees, Markov Chain Monte Carlo (MCMC) search algorithms are required to generate samples of tree models according to their posterior probabilities. The critical component of such an MCMC algorithm is to construct good Metropolis-Hastings steps for updating the tree topology. However, such algorithms frequently suffering from local mode stickiness and poor mixing. As a result, the algorithms are slow to converge. Hitherto, authors have primarily used discrete-time birth/death mechanisms for Bayesian (sums of) regression tree models to explore the model space. These algorithms are efficient only if the acceptance rate is high which is not always the case. Here we overcome this issue by developing a new search algorithm which is based on a continuous-time birth-death Markov process. This search algorithm explores the model space by jumping between parameter spaces corresponding to different tree structures. In the proposed algorithm, the moves between models are always accepted which can dramatically improve the convergence and mixing properties of the MCMC algorithm. We provide theoretical support of the algorithm for Bayesian regression tree models and demonstrate its performance.

In many cases an intelligent agent may want to learn how to mimic a single observed demonstrated trajectory. In this work we consider how to perform such procedural learning from observation, which could help to enable agents to better use the enormous set of video data on observation sequences. Our approach exploits the properties of this setting to incrementally build an open loop action plan that can yield the desired subsequence, and can be used in both Markov and partially observable Markov domains. In addition, procedures commonly involve repeated extended temporal action subsequences. Our method optimistically explores actions to leverage potential repeated structure in the procedure. In comparing to some state-of-the-art approaches we find that our explicit procedural learning from observation method is about 100 times faster than policy-gradient based approaches that learn a stochastic policy and is faster than model based approaches as well. We also find that performing optimistic action selection yields substantial speed ups when latent dynamical structure is present.

In many applications, observed data are influenced by some combination of latent causes. For example, suppose sensors are placed inside a building to record responses such as temperature, humidity, power consumption and noise levels. These random, observed responses are typically affected by many unobserved, latent factors (or features) within the building such as the number of individuals, the turning on and off of electrical devices, power surges, etc. These latent factors are usually present for a contiguous period of time before disappearing; further, multiple factors could be present at a time. This paper develops new probabilistic methodology and inference methods for random object generation influenced by latent features exhibiting temporal persistence. Every datum is associated with subsets of a potentially infinite number of hidden, persistent features that account for temporal dynamics in an observation. The ensuing class of dynamic models constructed by adapting the Indian Buffet Process --- a probability measure on the space of random, unbounded binary matrices --- finds use in a variety of applications arising in operations, signal processing, biomedicine, marketing, image analysis, etc. Illustrations using synthetic and real data are provided.

Predicting the outcomes of integrating Unmanned Aerial Systems (UAS) into the National Aerospace (NAS) is a complex problem which is required to be addressed by simulation studies before allowing the routine access of UAS into the NAS. This thesis focuses on providing 2D and 3D simulation frameworks using a game theoretical methodology to evaluate integration concepts in scenarios where manned and unmanned air vehicles co-exist. The fundamental gap in the literature is that the models of interaction between manned and unmanned vehicles are insufficient: a) they assume that pilot behavior is known a priori and b) they disregard decision making processes. The contribution of this work is to propose a modeling framework, in which, human pilot reactions are modeled using reinforcement learning and a game theoretical concept called level-k reasoning to fill this gap. The level-k reasoning concept is based on the assumption that humans have various levels of decision making. Reinforcement learning is a mathematical learning method that is rooted in human learning. In this work, a classical and an approximate reinforcement learning (Neural Fitted Q Iteration) methods are used to model time-extended decisions of pilots with 2D and 3D maneuvers. An analysis of UAS integration is conducted using example scenarios in the presence of manned aircraft and fully autonomous UAS equipped with sense and avoid algorithms.

The ability to use data about prior decisions and their outcomes to make counterfactual inferences about how alternative decision policies might perform, is a cornerstone of intelligent behavior. It also has immense practical potential - it can enable the use of electronic medical record data to infer better treatment decisions for patients, the use of prior product recommendations to inform more effective strategies for presenting recommendations, and previously collected data from students using educational software to better teach those and future students. Such counterfactual reasoning, particularly when one is deriving decision policies that will be used to make not one but a sequence of decisions, is important since online sampling during a learning procedure is both costly and dangerous, and not practical in many of the applications above. While amply motivated, doing such counterfactual reasoning is also challenging because the data is censored - we can only observe the result of providing a particular chemotherapy treatment policy to a particular patient, not the counterfactual of if we were then to start with a radiation sequence. We focus on the problem of performing such counterfactual inferences in the context of sequential decision making in a Markov decision process (MDP).