Humans rely on strong inductive biases to learn from few examples and abstract useful information from sensory data. Instilling such biases in machine learning models has been shown to improve their performance on various benchmarks including few-shot learning, robustness, and alignment. However, finding effective training procedures to achieve that goal can be challenging as psychologically-rich training data such as human similarity judgments are expensive to scale, and Bayesian models of human inductive biases are often intractable for complex, realistic domains. Here, we address this challenge by introducing a Bayesian notion of generative similarity whereby two datapoints are considered similar if they are likely to have been sampled from the same distribution. This measure can be applied to complex generative processes, including probabilistic programs. We show that generative similarity can be used to define a contrastive learning objective even when its exact form is intractable, enabling learning of spatial embeddings that express specific inductive biases. We demonstrate the utility of our approach by showing how it can be used to capture human inductive biases for geometric shapes, and to better distinguish different abstract drawing styles that are parameterized by probabilistic programs.

Federated Learning (FL) enables collaborative training of machine learning models on decentralized data while preserving data privacy. However, data across clients often differs significantly due to class imbalance, feature distribution skew, sample size imbalance, and other phenomena. Leveraging information from these not identically distributed (non-IID) datasets poses substantial challenges. FL methods based on a single global model cannot effectively capture the variations in client data and underperform in non-IID settings. Consequently, Personalized FL (PFL) approaches that adapt to each client's data distribution but leverage other clients' data are essential but currently underexplored. We propose a novel Bayesian PFL framework using bi-level optimization to tackle the data heterogeneity challenges. Our proposed framework utilizes the global model as a prior distribution within a Maximum A Posteriori (MAP) estimation of personalized client models. This approach facilitates PFL by integrating shared knowledge from the prior, thereby enhancing local model performance, generalization ability, and communication efficiency. We extensively evaluated our bi-level optimization approach on real-world and synthetic datasets, demonstrating significant improvements in model accuracy compared to existing methods while reducing communication overhead. This study contributes to PFL by establishing a solid theoretical foundation for the proposed method and offering a robust, ready-to-use framework that effectively addresses the challenges posed by non-IID data in FL.

Language generation based on maximum likelihood estimation (MLE) has become the fundamental approach for text generation. Maximum likelihood estimation is typically performed by minimizing the log-likelihood loss, also known as the logarithmic score in statistical decision theory. The logarithmic score is strictly proper in the sense that it encourages honest forecasts, where the expected score is maximized only when the model reports true probabilities. Although many strictly proper scoring rules exist, the logarithmic score is the only local scoring rule among them that depends exclusively on the probability of the observed sample, making it capable of handling the exponentially large sample space of natural text. In this work, we propose a straightforward strategy for adapting scoring rules to language generation, allowing for language modeling with any non-local scoring rules. Leveraging this strategy, we train language generation models using two classic strictly proper scoring rules, the Brier score and the Spherical score, as alternatives to the logarithmic score. Experimental results indicate that simply substituting the loss function, without adjusting other hyperparameters, can yield substantial improvements in model's generation capabilities. Moreover, these improvements can scale up to large language models (LLMs) such as LLaMA-7B and LLaMA-13B. Source code: \url{https://github.com/shaochenze/ScoringRulesLM}.

For aligning large language models (LLMs), prior work has leveraged reinforcement learning via human feedback (RLHF) or variations of direct preference optimization (DPO). While DPO offers a simpler framework based on maximum likelihood estimation, it compromises on the ability to tune language models to easily maximize non-differentiable and non-binary objectives according to the LLM designer's preferences (e.g., using simpler language or minimizing specific kinds of harmful content). These may neither align with user preferences nor even be able to be captured tractably by binary preference data. To leverage the simplicity and performance of DPO with the generalizability of RL, we propose a hybrid approach between DPO and RLHF. With a simple augmentation to the implicit reward decomposition of DPO, we allow for tuning LLMs to maximize a set of arbitrary auxiliary rewards using offline RL. The proposed method, Hybrid Preference Optimization (HPO), shows the ability to effectively generalize to both user preferences and auxiliary designer objectives, while preserving alignment performance across a range of challenging benchmarks and model sizes.

In the present era of deep learning, continual learning research is mainly focused on mitigating forgetting when training a neural network with stochastic gradient descent on a non-stationary stream of data. On the other hand, in the more classical literature of statistical machine learning, many models have sequential Bayesian update rules that yield the same learning outcome as the batch training, i.e., they are completely immune to catastrophic forgetting. However, they are often overly simple to model complex real-world data. In this work, we adopt the meta-learning paradigm to combine the strong representational power of neural networks and simple statistical models' robustness to forgetting. In our novel meta-continual learning framework, continual learning takes place only in statistical models via ideal sequential Bayesian update rules, while neural networks are meta-learned to bridge the raw data and the statistical models. Since the neural networks remain fixed during continual learning, they are protected from catastrophic forgetting. This approach not only achieves significantly improved performance but also exhibits excellent scalability. Since our approach is domain-agnostic and model-agnostic, it can be applied to a wide range of problems and easily integrated with existing model architectures.

In applied statistics and machine learning, the "gold standards" used for training are often biased and almost always noisy. Dawid and Skene's justifiably popular crowdsourcing model adjusts for rater (coder, annotator) sensitivity and specificity, but fails to capture distributional properties of rating data gathered for training, which in turn biases training. In this study, we introduce a general purpose measurement-error model with which we can infer consensus categories by adding item-level effects for difficulty, discriminativeness, and guessability. We further show how to constrain the bimodal posterior of these models to avoid (or if necessary, allow) adversarial raters. We validate our model's goodness of fit with posterior predictive checks, the Bayesian analogue of $\chi^2$ tests. Dawid and Skene's model is rejected by goodness of fit tests, whereas our new model, which adjusts for item heterogeneity, is not rejected. We illustrate our new model with two well-studied data sets, binary rating data for caries in dental X-rays and implication in natural language.

Semi-implicit variational inference (SIVI) extends traditional variational families with semi-implicit distributions defined in a hierarchical manner. Due to the intractable densities of semi-implicit distributions, classical SIVI often resorts to surrogates of evidence lower bound (ELBO) that would introduce biases for training. A recent advancement in SIVI, named SIVI-SM, utilizes an alternative score matching objective made tractable via a minimax formulation, albeit requiring an additional lower-level optimization. In this paper, we propose kernel SIVI (KSIVI), a variant of SIVI-SM that eliminates the need for lower-level optimization through kernel tricks. Specifically, we show that when optimizing over a reproducing kernel Hilbert space (RKHS), the lower-level problem has an explicit solution. This way, the upper-level objective becomes the kernel Stein discrepancy (KSD), which is readily computable for stochastic gradient descent due to the hierarchical structure of semi-implicit variational distributions. An upper bound for the variance of the Monte Carlo gradient estimators of the KSD objective is derived, which allows us to establish novel convergence guarantees of KSIVI. We demonstrate the effectiveness and efficiency of KSIVI on both synthetic distributions and a variety of real data Bayesian inference tasks.

Diffusion models (DMs) have recently shown outstanding capability in modeling complex image distributions, making them expressive image priors for solving Bayesian inverse problems. However, most existing DM-based methods rely on approximations in the generative process to be generic to different inverse problems, leading to inaccurate sample distributions that deviate from the target posterior defined within the Bayesian framework. To harness the generative power of DMs while avoiding such approximations, we propose a Markov chain Monte Carlo algorithm that performs posterior sampling for general inverse problems by reducing it to sampling the posterior of a Gaussian denoising problem. Crucially, we leverage a general DM formulation as a unified interface that allows for rigorously solving the denoising problem with a range of state-of-the-art DMs. We demonstrate the effectiveness of the proposed method on six inverse problems (three linear and three nonlinear), including a real-world black hole imaging problem. Experimental results indicate that our proposed method offers more accurate reconstructions and posterior estimation compared to existing DM-based imaging inverse methods.

This paper investigates sparse high-dimensional linear regression, particularly examining the properties of the posterior under conditions of random design and unknown error variance. We provide consistency results for the posterior and analyze its concentration rates, demonstrating adaptiveness to the unknown sparsity level of the regression coefficient vector. Furthermore, we extend our investigation to establish concentration outcomes for parameter estimation using specific distance measures. These findings are in line with recent discoveries in frequentist studies. Additionally, by employing techniques to address model misspecification through a fractional posterior, we broaden our analysis through oracle inequalities to encompass the critical aspect of model misspecification for the regular posterior. Our novel findings are demonstrated using two different types of sparsity priors: a shrinkage prior and a spike-and-slab prior.

Real-world decision-making requires grappling with a perpetual lack of data as environments change; intelligent agents must comprehend uncertainty and actively gather information to resolve it. We propose a new framework for learning bandit algorithms from massive historical data, which we demonstrate in a cold-start recommendation problem. First, we use historical data to pretrain an autoregressive model to predict a sequence of repeated feedback/rewards (e.g., responses to news articles shown to different users over time). In learning to make accurate predictions, the model implicitly learns an informed prior based on rich action features (e.g., article headlines) and how to sharpen beliefs as more rewards are gathered (e.g., clicks as each article is recommended). At decision-time, we autoregressively sample (impute) an imagined sequence of rewards for each action, and choose the action with the largest average imputed reward. Far from a heuristic, our approach is an implementation of Thompson sampling (with a learned prior), a prominent active exploration algorithm. We prove our pretraining loss directly controls online decision-making performance, and we demonstrate our framework on a news recommendation task where we integrate end-to-end fine-tuning of a pretrained language model to process news article headline text to improve performance.