The main goal of Few-Shot learning algorithms is to enable learning from small amounts of data. One of the most popular and elegant Few-Shot learning approaches is Model-Agnostic Meta-Learning (MAML). In this paper, we propose a novel framework for Bayesian MAML called BH-MAML, which employs Hypernetworks for weight updates. It learns the universal weights point-wise, but a probabilistic structure is added when adapted for specific tasks. In such a framework, we can use simple Gaussian distributions or more complicated posteriors induced by Continuous Normalizing Flows.

Clustering tabular data remains a significant open challenge in data analysis and machine learning. Unlike for image data, similarity between tabular records often varies across datasets, making the definition of clusters highly dataset-dependent. Furthermore, the absence of supervised signals complicates hyperparameter tuning in deep learning clustering methods, frequently resulting in unstable performance. To address these issues and minimize the need for per-dataset tuning, we adopt an emerging approach in deep learning: zero-shot learning. We propose ZEUS, a self-contained model capable of clustering new datasets without any additional training or fine-tuning. It operates by decomposing complex datasets into meaningful components that can then be clustered effectively. Thanks to pre-training on synthetic datasets generated from a latent-variable prior, it generalizes across various datasets without requiring user intervention. To the best of our knowledge, ZEUS is the first zero-shot method capable of generating embeddings for tabular data in a fully unsupervised manner. Experimental results demonstrate that it performs on par with or better than traditional clustering algorithms and recent deep learning-based methods, while being significantly faster and more user-friendly.

Gaussian Processes (GPs) provide a convenient framework for specifying function-space priors, making them a natural choice for modeling uncertainty. In contrast, Bayesian Neural Networks (BNNs) offer greater scalability and extendability but lack the advantageous properties of GPs. This motivates the development of BNNs capable of replicating GP-like behavior. However, existing solutions are either limited to specific GP kernels or rely on heuristics. We demonstrate that trainable activations are crucial for effective mapping of GP priors to wide BNNs. Specifically, we leverage the closed-form 2-Wasserstein distance for efficient gradient-based optimization of reparameterized priors and activations. Beyond learned activations, we also introduce trainable periodic activations that ensure global stationarity by design, and functional priors conditioned on GP hyperparameters to allow efficient model selection. Empirically, our method consistently outperforms existing approaches or matches performance of the heuristic methods, while offering stronger theoretical foundations.

The behavior of many Bayesian models used in machine learning critically depends on the choice of prior distributions, controlled by some hyperparameters typically selected through Bayesian optimization or cross-validation. This requires repeated, costly, posterior inference. We provide an alternative for selecting good priors without carrying out posterior inference, building on the prior predictive distribution that marginalizes the model parameters. We estimate virtual statistics for data generated by the prior predictive distribution and then optimize over the hyperparameters to learn those for which the virtual statistics match the target values provided by the user or estimated from (a subset of) the observed data. We apply the principle for probabilistic matrix factorization, for which good solutions for prior selection have been missing. We show that for Poisson factorization models we can analytically determine the hyperparameters, including the number of factors, that best replicate the target statistics, and we empirically study the sensitivity of the approach for the model mismatch. We also present a model-independent procedure that determines the hyperparameters for general models by stochastic optimization and demonstrate this extension in the context of hierarchical matrix factorization models.

Uplift modeling refers to estimating the causal effect of a treatment on an individual observation, used for instance to identify customers worth targeting with a discount in e-commerce. We introduce a simple yet effective undersampling strategy for dealing with the prevalent problem of high class imbalance (low conversion rate) in such applications. Our strategy is agnostic to the base learners and produces a 6.5% improvement over the best published benchmark for the largest public uplift data which incidentally exhibits high class imbalance. We also introduce a new metric on calibration for uplift modeling and present a strategy to improve the calibration of the proposed method.

Bayesian models quantify uncertainty and facilitate optimal decision-making in downstream applications. For most models, however, practitioners are forced to use approximate inference techniques that lead to sub-optimal decisions due to incorrect posterior predictive distributions. We present a novel approach that corrects for inaccuracies in posterior inference by altering the decision-making process. We train a separate model to make optimal decisions under the approximate posterior, combining interpretable Bayesian modeling with optimization of direct predictive accuracy in a principled fashion. The solution is generally applicable as a plug-in module for predictive decision-making for arbitrary probabilistic programs, irrespective of the posterior inference strategy. We demonstrate the approach empirically in several problems, confirming its potential.

Bayesian decision theory outlines a rigorous framework for making optimal decisions based on maximizing expected utility over a model posterior. However, practitioners often do not have access to the full posterior and resort to approximate inference strategies. In such cases, taking the eventual decision-making task into account while performing the inference allows for calibrating the posterior approximation to maximize the utility. We present an automatic pipeline that co-opts continuous utilities into variational inference algorithms to account for decision-making. We provide practical strategies for approximating and maximizing the gain, and empirically demonstrate consistent improvement when calibrating approximations for specific utilities.

Badges are a common, and sometimes the only, method of incentivizing users to perform certain actions on on- line sites. However, due to many competing factors influencing user temporal dynamics, it is difficult to determine whether the badge had (or will have) the intended effect or not. In this paper, we introduce two complementary approaches for determining badge influence on users. In the first one, we cluster users’ temporal traces (represented with Poisson processes) and apply covariates (user features) to regularize results. In the second approach, we first classify users’ temporal traces with a novel statistical framework, and then we refine the classification results with a semi-supervised clustering of covariates. Outcomes obtained from an evaluation on synthetic datasets and experiments on two badges from a pop- ular Q&A platform confirm that it is possible to validate, characterize and to some extent predict users affected by the badge.

A wide variety of online platforms use digital badges to encourage users to take certain types of desirable actions. However, despite their growing popularity, their causal effect on users» behavior is not well understood. This is partly due to the lack of counterfactual data and the myriad of complex factors that influence users» behavior over time. As a consequence, their design and deployment lacks general principles. In this paper, we focus on first-time badges, which are awarded after a user takes a particular type of action for the first time, and study their causal effect by harnessing the delayed introduction of several badges in a popular Q&A website. In doing so, we introduce a novel causal inference framework for first-time badges whose main technical innovations are a robust survival-based hypothesis testing procedure, which controls for the heterogeneity in the benefit users obtain from taking an action, and a bootstrap difference-in-differences method, which controls for the random fluctuations in users» behavior over time. Our results suggest that first-time badges steer users» behavior if the initial benefit a user obtains from taking the corresponding action is sufficiently low, otherwise, we do not find significant effects. Moreover, for badges that successfully steered user behavior, we perform a counterfactual analysis and show that they significantly improved the functioning of the site at a community level.

Dietary pattern analysis is an important research area, and recently the availability of rich resources in food-focused social networks has enabled new opportunities in that field. However, there is a little understanding of how online textual content is related to actual health factors, e.g., nutritional values. To contribute to this lack of knowledge, we present a novel approach to mine and model online food content by combining text topics with related nutrient facts. Our empirical analysis reveals a strong correlation between them and our experiments show the extent to which it is possible to predict nutrient facts from meal name.

Historically, there have always been differences in how men and women cook or eat. The reasons for this gender divide have mostly gone in Western culture, but still there is qualitative and anecdotal evidence that men prefer heftier food, that women take care of everyday cooking, and that men cook to impress. In this paper, we show that these differences can also quantitatively be observed in a large dataset of almost 200 thousand members of an online recipe community. Further, we show that, using a set of 88 features, the gender of the cooks can be predicted with fairly good accuracy of 75%, with preference for particular dishes, the use of spices and the use of kitchen utensils being the strongest predictors. Finally, we show the positive impact of our results on online food recipe recommender systems that take gender information into account.