For many years, the two dominant paradigms in artificial intelligence (AI) have been logical AI and statistical AI. Logical AI uses first-order logic and related representations to capture complex relationships and knowledge about the world. However, logic-based approaches are often too brittle to handle the uncertainty and noise present in many applications. Statistical AI uses probabilistic representations such as probabilistic graphical models to capture uncertainty. However, graphical models only represent distributions over propositional universes and must be customized to handle relational domains. As a result, expressing complex concepts and relationships in graphical models is often difficult and labor-intensive.

Sequential decision making in the presence of uncertainty and stochastic dynamics gives rise to distributions over state/action trajectories in reinforcement learning (RL) and optimal control problems. This observation has led to a variety of connections between RL and inference in probabilistic graphical models (PGMs). Here we explore a different dimension to this relationship, examining reinforcement learning using the tools and abstractions of statistical physics. The central object in the statistical physics abstraction is the idea of a partition function $\mathcal{Z}$, and here we construct a partition function from the ensemble of possible trajectories that an agent might take in a Markov decision process. Although value functions and $Q$-functions can be derived from this partition function and interpreted via average energies, the $\mathcal{Z}$-function provides an object with its own Bellman equation that can form the basis of alternative dynamic programming approaches. Moreover, when the MDP dynamics are deterministic, the Bellman equation for $\mathcal{Z}$ is linear, allowing direct solutions that are unavailable for the nonlinear equations associated with traditional value functions. The policies learned via these $\mathcal{Z}$-based Bellman updates are tightly linked to Boltzmann-like policy parameterizations. In addition to sampling actions proportionally to the exponential of the expected cumulative reward as Boltzmann policies would, these policies take entropy into account favoring states from which many outcomes are possible.

In Reinforcement Learning we look for meaning in the flow of input/output information. If we do not find meaning, the information flow is not more than noise to us. Before we are able to find meaning, we should first learn how to discover and identify objects. What is an object? In this article we will demonstrate that an object is an event-driven model. These models are a generalization of action-driven models. In Markov Decision Process we have an action-driven model which changes its state at each step. The advantage of event-driven models is their greater sustainability as they change their states only upon the occurrence of particular events. These events may occur very rarely, therefore the state of the event-driven model is much more predictable.

The work presented in this master thesis consists of extracting a set of events from texts written in natural language. For this purpose, we have based ourselves on the basic notions of the information extraction as well as the open information extraction. First, we applied an open information extraction(OIE) system for the relationship extraction, to highlight the importance of OIEs in event extraction, and we used the ontology to the event modeling. We tested the results of our approach with test metrics. As a result, the two-level event extraction approach has shown good performance results but requires a lot of expert intervention in the construction of classifiers and this will take time. In this context we have proposed an approach that reduces the expert intervention in the relation extraction, the recognition of entities and the reasoning which are automatic and based on techniques of adaptation and correspondence. Finally, to prove the relevance of the extracted results, we conducted a set of experiments using different test metrics as well as a comparative study.

When agents interact with a complex environment, they must form and maintain beliefs about the relevant aspects of that environment. We propose a way to efficiently train expressive generative models in complex environments. We show that a predictive algorithm with an expressive generative model can form stable belief-states in visually rich and dynamic 3D environments. More precisely, we show that the learned representation captures the layout of the environment as well as the position and orientation of the agent. Our experiments show that the model substantially improves data-efficiency on a number of reinforcement learning (RL) tasks compared with strong model-free baseline agents. We find that predicting multiple steps into the future (overshooting), in combination with an expressive generative model, is critical for stable representations to emerge. In practice, using expressive generative models in RL is computationally expensive and we propose a scheme to reduce this computational burden, allowing us to build agents that are competitive with model-free baselines.

Computer science as an academic discipline began in the 1960s. Emphasis was on programming languages, compilers, operating systems, and the mathematical theory that supported these areas. Courses in theoretical computer science covered finite automata, regular expressions, context-free languages, and computability. In the 1970s, the study of algorithms was added as an important component of theory. The emphasis was on making computers useful.

We consider the problem of imitation learning from expert demonstrations in partially observable Markov decision processes (POMDPs). Belief representations, which characterize the distribution over the latent states in a POMDP, have been modeled using recurrent neural networks and probabilistic latent variable models, and shown to be effective for reinforcement learning in POMDPs. In this work, we investigate the belief representation learning problem for generative adversarial imitation learning in POMDPs. Instead of training the belief module and the policy separately as suggested in prior work, we learn the belief module jointly with the policy, using a task-aware imitation loss to ensure that the representation is more aligned with the policy's objective. To improve robustness of representation, we introduce several informative belief regularization techniques, including multi-step prediction of dynamics and action-sequences. Evaluated on various partially observable continuous-control locomotion tasks, our belief-module imitation learning approach (BMIL) substantially outperforms several baselines, including the original GAIL algorithm and the task-agnostic belief learning algorithm. Extensive ablation analysis indicates the effectiveness of task-aware belief learning and belief regularization.

Animals need to devise strategies to maximize returns while interacting with their environment based on incoming noisy sensory observations. Task-relevant states, such as the agent's location within an environment or the presence of a predator, are often not directly observable but must be inferred using available sensory information. Successor representations (SR) have been proposed as a middle-ground between model-based and model-free reinforcement learning strategies, allowing for fast value computation and rapid adaptation to changes in the reward function or goal locations. Indeed, recent studies suggest that features of neural responses are consistent with the SR framework. However, it is not clear how such representations might be learned and computed in partially observed, noisy environments. Here, we introduce a neurally plausible model using distributional successor features, which builds on the distributed distributional code for the representation and computation of uncertainty, and which allows for efficient value function computation in partially observed environments via the successor representation. We show that distributional successor features can support reinforcement learning in noisy environments in which direct learning of successful policies is infeasible.

Traditional automatic speech recognition (ASR) systems often use an acoustic model (AM) built on handcrafted acoustic features, such as log Mel-filter bank (FBANK) values. Recent studies found that AMs with convolutional neural networks (CNNs) can directly use the raw waveform signal as input. Given sufficient training data, these AMs can yield a competitive word error rate (WER) to those built on FBANK features. This paper proposes a novel multi-span structure for acoustic modelling based on the raw waveform with multiple streams of CNN input layers, each processing a different span of the raw waveform signal. Evaluation on both the single channel CHiME4 and AMI data sets show that multi-span AMs give a lower WER than FBANK AMs by an average of about 5% (relative). Analysis of the trained multi-span model reveals that the CNNs can learn filters that are rather different to the log Mel filters. Furthermore, the paper shows that a widely used single span raw waveform AM can be improved by using a smaller CNN kernel size and increased stride to yield improved WERs.

Ensemble learning, the machine learning paradigm where multiple algorithms are combined, has exhibited promising perfomance in a variety of tasks. The present work focuses on unsupervised ensemble classification. The term unsupervised refers to the ensemble combiner who has no knowledge of the ground-truth labels that each classifier has been trained on. While most prior works on unsupervised ensemble classification are designed for independent and identically distributed (i.i.d.) data, the present work introduces an unsupervised scheme for learning from ensembles of classifiers in the presence of data dependencies. Two types of data dependencies are considered: sequential data and networked data whose dependencies are captured by a graph. Moment matching and Expectation Maximization algorithms are developed for the aforementioned cases, and their performance is evaluated on synthetic and real datasets.