Disentangled representation learning aims to represent the underlying generative factors of a dataset in a latent representation independently of one another. In our work, we propose a discrete variational autoencoder (VAE) based model where the ground truth information about the generative factors are not provided to the model. We demonstrate the advantages of learning discrete representations over learning continuous representations in facilitating disentanglement. Furthermore, we propose incorporating an inductive bias into the model to further enhance disentanglement. Precisely, we propose scalar quantization of the latent variables in a latent representation with scalar values from a global codebook, and we add a total correlation term to the optimization as an inductive bias. Our method called FactorQVAE combines optimization based disentanglement approaches with discrete representation learning, and it outperforms the former disentanglement methods in terms of two disentanglement metrics (DCI and InfoMEC) while improving the reconstruction performance. Our code can be found at https://github.com/ituvisionlab/FactorQVAE .

Reinforcement learning (RL) for continuous control under delayed rewards is an under-explored problem despite its significance in real-world applications. Many complex skills are based on intermediate ones as prerequisites. For instance, a humanoid locomotor must learn how to stand before it can learn to walk. To cope with delayed reward, an agent must perform deep exploration. However, existing deep exploration methods are designed for small discrete action spaces, and their generalization to state-of-the-art continuous control remains unproven. We address the deep exploration problem for the first time from a PAC-Bayesian perspective in the context of actor-critic learning. To do this, we quantify the error of the Bellman operator through a PAC-Bayes bound, where a bootstrapped ensemble of critic networks represents the posterior distribution, and their targets serve as a data-informed function-space prior. We derive an objective function from this bound and use it to train the critic ensemble. Each critic trains an individual soft actor network, implemented as a shared trunk and critic-specific heads. The agent performs deep exploration by acting epsilon-softly on a randomly chosen actor head. Our proposed algorithm, named PAC-Bayesian Actor-Critic (PBAC), is the only algorithm to consistently discover delayed rewards on continuous control tasks with varying difficulty.

Off-policy actor-critic algorithms have shown strong potential in deep reinforcement learning for continuous control tasks. Their success primarily comes from leveraging pessimistic state-action value function updates, which reduce function approximation errors and stabilize learning. However, excessive pessimism can limit exploration, preventing the agent from effectively refining its policies. Conversely, optimism can encourage exploration but may lead to high-risk behaviors and unstable learning if not carefully managed. To address this trade-off, we propose Utility Soft Actor-Critic (USAC), a novel framework that allows independent, interpretable control of pessimism and optimism for both the actor and the critic. USAC dynamically adapts its exploration strategy based on the uncertainty of critics using a utility function, enabling a task-specific balance between optimism and pessimism. This approach goes beyond binary choices of pessimism or optimism, making the method both theoretically meaningful and practically feasible. Experiments across a variety of continuous control tasks show that adjusting the degree of pessimism or optimism significantly impacts performance. When configured appropriately, USAC consistently outperforms state-of-the-art algorithms, demonstrating its practical utility and feasibility.

Current approaches to model-based offline reinforcement learning often incorporate uncertainty-based reward penalization to address the distributional shift problem. These approaches, commonly known as pessimistic value iteration, use Monte Carlo sampling to estimate the Bellman target to perform temporal difference-based policy evaluation. We find out that the randomness caused by this sampling step significantly delays convergence. We present a theoretical result demonstrating the strong dependency of suboptimality on the number of Monte Carlo samples taken per Bellman target calculation. Our main contribution is a deterministic approximation to the Bellman target that uses progressive moment matching, a method developed originally for deterministic variational inference. The resulting algorithm, which we call Moment Matching Offline Model-Based Policy Optimization (MOMBO), propagates the uncertainty of the next state through a nonlinear Q-network in a deterministic fashion by approximating the distributions of hidden layer activations by a normal distribution. We show that it is possible to provide tighter guarantees for the suboptimality of MOMBO than the existing Monte Carlo sampling approaches. We also observe MOMBO to converge faster than these approaches in a large set of benchmark tasks.

We present improved algorithms with worst-case regret guarantees for the stochastic linear bandit problem. The widely used "optimism in the face of uncertainty" principle reduces a stochastic bandit problem to the construction of a confidence sequence for the unknown reward function. The performance of the resulting bandit algorithm depends on the size of the confidence sequence, with smaller confidence sets yielding better empirical performance and stronger regret guarantees. In this work, we use a novel tail bound for adaptive martingale mixtures to construct confidence sequences which are suitable for stochastic bandits. These confidence sequences allow for efficient action selection via convex programming. We prove that a linear bandit algorithm based on our confidence sequences is guaranteed to achieve competitive worst-case regret. We show that our confidence sequences are tighter than competitors, both empirically and theoretically. Finally, we demonstrate that our tighter confidence sequences give improved performance in several hyperparameter tuning tasks.

PAC-Bayes has recently re-emerged as an effective theory with which one can derive principled learning algorithms with tight performance guarantees. However, applications of PAC-Bayes to bandit problems are relatively rare, which is a great misfortune. Many decision-making problems in healthcare, finance and natural sciences can be modelled as bandit problems. In many of these applications, principled algorithms with strong performance guarantees would be very much appreciated. This survey provides an overview of PAC-Bayes bounds for bandit problems and an experimental comparison of these bounds. On the one hand, we found that PAC-Bayes bounds are a useful tool for designing offline bandit algorithms with performance guarantees. In our experiments, a PAC-Bayesian offline contextual bandit algorithm was able to learn randomised neural network polices with competitive expected reward and non-vacuous performance guarantees. On the other hand, the PAC-Bayesian online bandit algorithms that we tested had loose cumulative regret bounds. We conclude by discussing some topics for future work on PAC-Bayesian bandit algorithms.

Neural Stochastic Differential Equations (NSDEs) model the drift and diffusion functions of a stochastic process as neural networks. While NSDEs are known to make accurate predictions, their uncertainty quantification properties have been remained unexplored so far. We report the empirical finding that obtaining well-calibrated uncertainty estimations from NSDEs is computationally prohibitive. As a remedy, we develop a computationally affordable deterministic scheme which accurately approximates the transition kernel, when dynamics is governed by a NSDE. Our method introduces a bidimensional moment matching algorithm: vertical along the neural net layers and horizontal along the time direction, which benefits from an original combination of effective approximations. Our deterministic approximation of the transition kernel is applicable to both training and prediction. We observe in multiple experiments that the uncertainty calibration quality of our method can be matched by Monte Carlo sampling only after introducing high computational cost. Thanks to the numerical stability of deterministic training, our method also improves prediction accuracy.

We study uncertainty-aware modeling of continuous-time dynamics of interacting objects. We introduce a new model that decomposes independent dynamics of single objects accurately from their interactions. By employing latent Gaussian process ordinary differential equations, our model infers both independent dynamics and their interactions with reliable uncertainty estimates. In our formulation, each object is represented as a graph node and interactions are modeled by accumulating the messages coming from neighboring objects. We show that efficient inference of such a complex network of variables is possible with modern variational sparse Gaussian process inference techniques. We empirically demonstrate that our model improves the reliability of long-term predictions over neural network based alternatives and it successfully handles missing dynamic or static information. Furthermore, we observe that only our model can successfully encapsulate independent dynamics and interaction information in distinct functions and show the benefit from this disentanglement in extrapolation scenarios.

Neural Stochastic Differential Equations model a dynamical environment with neural nets assigned to their drift and diffusion terms. The high expressive power of their nonlinearity comes at the expense of instability in the identification of the large set of free parameters. This paper presents a recipe to improve the prediction accuracy of such models in three steps: i) accounting for epistemic uncertainty by assuming probabilistic weights, ii) incorporation of partial knowledge on the state dynamics, and iii) training the resultant hybrid model by an objective derived from a PAC-Bayesian generalization bound. We observe in our experiments that this recipe effectively translates partial and noisy prior knowledge into an improved model fit.

Model selection is treated as a standard performance boosting step in many machine learning applications. Once all other properties of a learning problem are fixed, the model is selected by grid search on a held-out validation set. This is strictly inapplicable to active learning. Within the standardized workflow, the acquisition function is chosen among available heuristics a priori, and its success is observed only after the labeling budget is already exhausted. More importantly, none of the earlier studies report a unique consistently successful acquisition heuristic to the extent to stand out as the unique best choice. We present a method to break this vicious circle by defining the acquisition function as a learning predictor and training it by reinforcement feedback collected from each labeling round. As active learning is a scarce data regime, we bootstrap from a well-known heuristic that filters the bulk of data points on which all heuristics would agree, and learn a policy to warp the top portion of this ranking in the most beneficial way for the character of a specific data distribution. Our system consists of a Bayesian neural net, the predictor, a bootstrap acquisition function, a probabilistic state definition, and another Bayesian policy network that can effectively incorporate this input distribution. We observe on three benchmark data sets that our method always manages to either invent a new superior acquisition function or to adapt itself to the a priori unknown best performing heuristic for each specific data set.

Distributed Denial of Service (DDoS) attacks continue to be one of the most severe threats in the Internet. The intrinsic challenge in preventing DDoS attacks is to distinguish them from legitimate flash crowds since two have many traffic characteristics in common. Today most DDoS detection techniques focus on finding parametric differences between the patterns in attack and legitimate traffic. However, such techniques are very sensitive to the threshold values set on the parameters and more importantly legitimate traffic features might be mimicked by smart attackers to generate requests that look like flash crowds. In this paper, we propose a framework for training networks for such smart attacks. Our framework is based on Deep Generative Network models and our contributions are two-fold.We first show that legitimate traffic features can be mimicked without explicitly modeling their distributions. Second, we introduce the concept of context-aware DDoS attacks. We show that an attacker can generate traffic that looks similar to flash crowds to be undetected for long periods of time. However, the ability of generating such attacks is constrained by the budget of the attacker. A context-aware attacker is the one that can intelligently use its budget to maximize the damage in the victim network. Our study provides a framework for training networks for such DDoS attack scenarios.

Deterministic neural nets have been shown to learn effective predictors on a wide range of machine learning problems. However, as the standard approach is to train the network to minimize a prediction loss, the resultant model remains ignorant to its prediction confidence. Orthogonally to Bayesian neural nets that indirectly infer prediction uncertainty through weight uncertainties, we propose explicit modeling of the same using the theory of subjective logic. By placing a Dirichlet distribution on the class probabilities, we treat predictions of a neural net as subjective opinions and learn the function that collects the evidence leading to these opinions by a deterministic neural net from data. The resultant predictor for a multi-class classification problem is another Dirichlet distribution whose parameters are set by the continuous output of a neural net. We provide a preliminary analysis on how the peculiarities of our new loss function drive improved uncertainty estimation. We observe that our method achieves unprecedented success on detection of out-of-distribution queries and endurance against adversarial perturbations.

Gaussian Processes (GPs) are effective Bayesian predictors. We here show for the first time that instance labels of a GP classifier can be inferred in the multiple instance learning (MIL) setting using variational Bayes. We achieve this via a new construction of the bag likelihood that assumes a large value if the instance predictions obey the MIL constraints and a small value otherwise. This construction lets us derive the update rules for the variational parameters analytically, assuring both scalable learning and fast convergence. We observe this model to improve the state of the art in instance label prediction from bag-level supervision in the 20 Newsgroups benchmark, as well as in Barretts cancer tumor localization from histopathology tissue microarray images. Furthermore, we introduce a novel pipeline for weakly supervised object detection naturally complemented with our model, which improves the state of the art on the PASCAL VOC 2007 and 2012 data sets. Last but not least, the performance of our model can be further boosted up using mixed supervision: a combination of weak (bag) and strong (instance) labels.

We introduce a novel Gaussian process based Bayesian model for asymmetric transfer learning. We adopt a two-layer feed-forward deep Gaussian process as the task learner of source and target domains. The first layer projects the data onto a separate non-linear manifold for each task. We perform knowledge transfer by projecting the target data also onto the source domain and linearly combining its representations on the source and target domain manifolds. Our approach achieves the state-of-the-art in a benchmark real-world image categorization task, and improves on it in cross-tissue tumor detection from histopathology tissue slide images.

We introduce a multiple instance learning algorithm based on randomized decision trees. Our model extends an existing algorithm by Bloc keel et al. [2] in several ways: 1) We learn a random forest instead of a single tree. 2) We construct the trees by splits based on non-linear boundaries on multiple features at a time. 3) We learn an optimal way of combining the decisions of multiple trees under the multiple instance constraints (i.e. positive bags have at least one positive instance, negative bags have only negative instances). Experiments on the typical benchmark data sets show that this model's prediction performance is clearly better than earlier tree based methods, and is comparable to the global state-of-the-art.

We study the feasibility of the following idea: Could a system learn to use the user's natural eye movements to infer relevance of real-world objects, if the user produced a set of learning data by clicking a "relevance" button during a learning session? If the answer is yes, the combination of eye tracking and machine learning would give a basis of "natural" interaction with the system by normally looking around, which would be very useful in mobile proactive setups. We measured the eye movements of the users while they were exploring an artificial art gallery. They labeled the relevant paintings by clicking a button while looking at them. The results show that a Gaussian process classifier accompanied by a time series kernel on the eye movements within an object predicts whether that object is relevant with better accuracy than dwell-time thresholding and random guessing.

As prototypes of data glasses having both data augmentation and gaze tracking capabilities are becoming available, it is now possible to develop proactive gaze-controlled user interfaces to display information about objects, people, and other entities in real-world setups. In order to decide which objects the augmented information should be about, and how saliently to augment, the system needs an estimate of the importance or relevance of the objects of the scene for the user at a given time. The estimates will be used to minimize distraction of the user, and for providing efficient spatial management of the augmented items. This work is a feasibility study on inferring the relevance of objects in dynamic scenes from gaze. We collected gaze data from subjects watching a video for a pre-defined task. The results show that a simple ordinal logistic regression model gives relevance rankings of scene objects with a promising accuracy.