To offer accurate and diverse recommendation services, recent methods use auxiliary information to foster the learning process of user and item representations. Many SOTA methods fuse different sources of information (user, item, knowledge graph, tags, etc.) into a graph and use Graph Neural Networks to introduce the auxiliary information through the message passing paradigm. In this work, we seek an alternative framework that is light and effective through self-supervised learning across different sources of information, particularly for the commonly accessible item tag information. We use a self-supervision signal to pair users with the auxiliary information associated with the items they have interacted with before. To achieve the pairing, we create a proxy training task. For a given item, the model predicts the correct pairing between the representations obtained from the users that have interacted with this item and the assigned tags. This design provides an efficient solution, using the auxiliary information directly to enhance the quality of user and item embeddings. User behavior in recommendation systems is driven by the complex interactions of many factors behind the decision-making processes. To make the pairing process more fine-grained and avoid embedding collapse, we propose an intent-aware self-supervised pairing process where we split the user embeddings into multiple sub-embedding vectors. Each sub-embedding vector captures a specific user intent via self-supervised alignment with a particular cluster of tags. We integrate our designed framework with various recommendation models, demonstrating its flexibility and compatibility. Through comparison with numerous SOTA methods on seven real-world datasets, we show that our method can achieve better performance while requiring less training time. This indicates the potential of applying our approach on web-scale datasets.

Epistemic uncertainty quantification is a crucial part of drawing credible conclusions from predictive models, whether concerned about the prediction at a given point or any downstream evaluation that uses the model as input. When the predictive model is simple and its evaluation differentiable, this task is solved by the delta method, where we propagate the asymptotically-normal uncertainty in the predictive model through the evaluation to compute standard errors and Wald confidence intervals. However, this becomes difficult when the model and/or evaluation becomes more complex. Remedies include the bootstrap, but it can be computationally infeasible when training the model even once is costly. In this paper, we propose an alternative, the implicit delta method, which works by infinitesimally regularizing the training loss of the predictive model to automatically assess downstream uncertainty. We show that the change in the evaluation due to regularization is consistent for the asymptotic variance of the evaluation estimator, even when the infinitesimal change is approximated by a finite difference. This provides both a reliable quantification of uncertainty in terms of standard errors as well as permits the construction of calibrated confidence intervals. We discuss connections to other approaches to uncertainty quantification, both Bayesian and frequentist, and demonstrate our approach empirically.

For non-Gaussian (non-conjugate) likelihoods, two common choices for approximate inference are Expectation Propagation (EP) and Variational Inference (VI), which have complementary strengths and weaknesses. While VI's lower bound to the marginal likelihood is a suitable objective for inferring the approximate posterior, it does not automatically imply it is a good learning objective for hyperparameter optimization. We design a hybrid training procedure where the inference leverages conjugate-computation VI and the learning uses an EP-like marginal likelihood approximation. We empirically demonstrate on binary classification that this provides a good learning objective and generalizes better.

Kernel methods provide a flexible and theoretically grounded approach to nonlinear and nonparametric learning. While memory requirements hinder their applicability to large datasets, many approximate solvers were recently developed for scaling up kernel methods, such as random Fourier features. However, these scalable approaches are based on approximations of isotropic kernels, which are incapable of removing the influence of possibly irrelevant features. In this work, we design random Fourier features for automatic relevance determination kernels, widely used for variable selection, and propose a new method based on joint optimization of the kernel machine parameters and the kernel relevances. Additionally, we present a new optimization algorithm that efficiently tackles the resulting objective function, which is non-convex. Numerical validation on synthetic and real-world data shows that our approach achieves low prediction error and effectively identifies relevant predictors. Our solution is modular and uses the PyTorch framework.

We introduce a family of interpretable machine learning models, with two broad additions: Linearised Additive Models (LAMs) which replace the ubiquitous logistic link function in General Additive Models (GAMs); and SubscaleHedge, an expert advice algorithm for combining base models trained on subsets of features called subscales. LAMs can augment any additive binary classification model equipped with a sigmoid link function. Moreover, they afford direct global and local attributions of additive components to the model output in probability space. We argue that LAMs and SubscaleHedge improve the interpretability of their base algorithms. Using rigorous null-hypothesis significance testing on a broad suite of financial modelling data, we show that our algorithms do not suffer from large performance penalties in terms of ROC-AUC and calibration.

The fact that we can build models from data, and therefore refine our models with more data from experiments, is usually given for granted in scientific inquiry. However, how much information can we extract, and how precise can we expect our learned model to be, if we have only a finite amount of data at our disposal? Nuclear physics demands an high degree of precision from models that are inferred from the limited number of nuclei that can be possibly made in the laboratories. In manuscript I will introduce some concepts of computational science, such as statistical theory of learning and Hamiltonian complexity, and use them to contextualise the results concerning the amount of data necessary to extrapolate a mass model to a given precision.

Self-supervised speech representation learning aims to extract meaningful factors from the speech signal that can later be used across different downstream tasks, such as speech and/or emotion recognition. Existing models, such as HuBERT, however, can be fairly large thus may not be suitable for edge speech applications. Moreover, realistic applications typically involve speech corrupted by noise and room reverberation, hence models need to provide representations that are robust to such environmental factors. In this study, we build on the so-called DistilHuBERT model, which distils HuBERT to a fraction of its original size, with three modifications, namely: (i) augment the training data with noise and reverberation, while the student model needs to distill the clean representations from the teacher model; (ii) introduce a curriculum learning approach where increasing levels of noise are introduced as the model trains, thus helping with convergence and with the creation of more robust representations; and (iii) introduce a multi-task learning approach where the model also reconstructs the clean waveform jointly with the distillation task, thus also acting as an enhancement step to ensure additional environment robustness to the representation. Experiments on three SUPERB tasks show the advantages of the proposed method not only relative to the original DistilHuBERT, but also to the original HuBERT, thus showing the advantages of the proposed method for ``in the wild'' edge speech applications.

Uncertainty quantification (UQ) has increasing importance in building robust high-performance and generalizable materials property prediction models. It can also be used in active learning to train better models by focusing on getting new training data from uncertain regions. There are several categories of UQ methods each considering different types of uncertainty sources. Here we conduct a comprehensive evaluation on the UQ methods for graph neural network based materials property prediction and evaluate how they truly reflect the uncertainty that we want in error bound estimation or active learning. Our experimental results over four crystal materials datasets (including formation energy, adsorption energy, total energy, and band gap properties) show that the popular ensemble methods for uncertainty estimation is NOT the best choice for UQ in materials property prediction. For the convenience of the community, all the source code and data sets can be accessed freely at \url{https://github.com/usccolumbia/materialsUQ}.

We consider dynamic multi-product pricing and assortment problems under an unknown demand over T periods, where in each period, the seller decides on the price for each product or the assortment of products to offer to a customer who chooses according to an unknown Multinomial Logit Model (MNL). Such problems arise in many applications, including online retail and advertising. We propose a randomized dynamic pricing policy based on a variant of the Online Newton Step algorithm (ONS) that achieves a $O(d\sqrt{T}\log(T))$ regret guarantee under an adversarial arrival model. We also present a new optimistic algorithm for the adversarial MNL contextual bandits problem, which achieves a better dependency than the state-of-the-art algorithms in a problem-dependent constant $\kappa_2$ (potentially exponentially small). Our regret upper bound scales as $\tilde{O}(d\sqrt{\kappa_2 T}+ \log(T)/\kappa_2)$, which gives a stronger bound than the existing $\tilde{O}(d\sqrt{T}/\kappa_2)$ guarantees.

This paper focuses on stochastic saddle point problems with decision-dependent distributions. These are problems whose objective is the expected value of a stochastic payoff function and whose data distribution drifts in response to decision variables--a phenomenon represented by a distributional map. A common approach to accommodating distributional shift is to retrain optimal decisions once a new distribution is revealed, or repeated retraining. We introduce the notion of equilibrium points, which are the fixed points of this repeated retraining procedure, and provide sufficient conditions for their existence and uniqueness. To find equilibrium points, we develop deterministic and stochastic primal-dual algorithms and demonstrate their convergence with constant step-size in the former and polynomial decay step-size schedule in the latter. By modeling errors emerging from a stochastic gradient estimator as sub-Weibull random variables, we provide error bounds in expectation and in high probability that hold for each iteration. Without additional knowledge of the distributional map, computing saddle points is intractable. Thus we propose a condition on the distributional map--which we call opposing mixture dominance--that ensures that the objective is strongly-convex-strongly-concave. Finally, we demonstrate that derivative-free algorithms with a single function evaluation are capable of approximating saddle points