Adaptive network sparsification improves model compactness and accuracy.
problem Suboptimal network sparsification due to input-independent dropout.
method Dependent variational beta-Bernoulli dropout.
result Significantly more compact networks with consistent accuracy improvements.
Proposes a non-parametric method for deep discrete latent variable models.
problem Learning sparse discrete latent representations in deep models.
method Iterative algorithm with Beta-Bernoulli process prior and local data scaling.
result Improves sparsity and scalability of deep discrete latent variable models.
The beta-Bernoulli process provides a Bayesian nonparametric prior for models involving collections of binary-valued features. A draw from the beta process yields an infinite collection of probabilities in the unit interval, and a draw from the Bernoulli process turns these into binary-valued features. Recent work has …
A new parallel MCMC method for Indian Buffet Process models.
problem Slow inference in Indian Buffet Process models.
method Hybrid sampler combining collapsed and uncollapsed MCMC for parallel computation.
result Asymptotically exact parallel inference for Indian Buffet Process models.
A beta-negative binomial (BNB) process is proposed, leading to a beta-gamma-Poisson process, which may be viewed as a "multi-scoop" generalization of the beta-Bernoulli process. The BNB process is augmented into a beta-gamma-gamma-Poisson hierarchical structure, and applied as a nonparametric Bayesian prior for an infi…
New models discover new topics over time in topic modeling.
problem Discovering new topics over time in topic modeling.
method Nonparametric Bayesian models and Hungarian matching algorithm.
result Significantly faster than existing methods, discovering new topics in large datasets.
Stochastic variational inference (SVI) is emerging as the most promising candidate for scaling inference in Bayesian probabilistic models to large datasets. However, the performance of these methods has been assessed primarily in the context of Bayesian topic models, particularly latent Dirichlet allocation (LDA). Deri…
Super-resolution methods form high-resolution images from low-resolution images. In this paper, we develop a new Bayesian nonparametric model for super-resolution. Our method uses a beta-Bernoulli process to learn a set of recurring visual patterns, called dictionary elements, from the data. Because it is nonparametric…
New algorithm speeds Bayesian nonparametric model inference.
problem Slow inference in Bayesian nonparametric models.
method Decompose random measures into finite and infinite sub-measures; use different algorithms for each.
result Hybrid algorithm improves scalability and mixing.
Efficient CVI for NGFA improves GFA inference for large-scale data.
problem Inference limitations in GFA models for large-scale data.
method Collapsed variational inference for nonparametric Bayesian GFA.
result CVI algorithm effectively approximates NGFA posterior in collapsed space.
We characterize the class of exchangeable feature allocations assigning probability Vn,k∏l=1kWmlUn−ml to a feature allocation of n individuals, displaying k features with counts (m1,…,mk) for these features. Each element of this class is parametrized by a countable matrix V…
Paper tackles deep learning confounding factors, learns unseen factors.
problem Learning from data with unknown and potentially infinite confounding factors.
method Combines deep generative models with Bayesian non-parametric factor models (Indian Buffet Process).
result Model can learn from data with unknown and potentially infinite confounding factors.
Modeling correlated mutations in cancer for personalized treatment.
problem Identifying mutations for personalized cancer therapy in heterogeneous profiles.
method Proposed correlated zero-inflated negative binomial process with mixed beta-Bernoulli and variational inference.
result Identified biologically relevant correlations between somatic mutations.
While most Bayesian nonparametric models in machine learning have focused on the Dirichlet process, the beta process, or their variants, the gamma process has recently emerged as a useful nonparametric prior in its own right. Current inference schemes for models involving the gamma process are restricted to MCMC-based …
PPT optimizes transformer behavior by steering its latent posterior using prior samples.
problem Eliciting desired behavior from transformers without backpropagation.
method Posterior Prefix Tuning (PPT) uses predictive Monte Carlo (PMC) samples and importance sampling to optimize the latent posterior.
result PPT optimizes transformer behavior without backpropagation, achieving high utility across different utility functions.
AutoSeM automatically selects and balances auxiliary tasks in MTL.
problem Choosing and balancing auxiliary tasks in MTL.
method AutoSeM uses a Beta-Bernoulli multi-armed bandit with Thompson Sampling for task selection and a Gaussian Process for learning the mixing ratio.
result AutoSeM achieves significant performance boosts on GLUE language understanding tasks.
The paper improves theoretical guarantees for Thompson Sampling in cascading bandits.
problem Optimizing online recommender systems with cascading bandits.
method Develops and analyzes new Thompson Sampling algorithms for cascading bandits.
result Establishes the first theoretical guarantees on Thompson Sampling for cascading bandits.
Bayesian model identifies cancer pathways using genomic data.
problem Identifying altered pathways associated with specific cancer types.
method Bayesian semi-nonnegative tri-matrix factorization incorporating biological prior knowledge.
result Pathways identified can be used as prognostic biomarkers.
Conjugate pairs of distributions over infinite dimensional spaces are prominent in statistical learning theory, particularly due to the widespread adoption of Bayesian nonparametric methodologies for a host of models and applications. Much of the existing literature in the learning community focuses on processes posses…
This paper introduces a Bayesian framework for optimizing online experiments to maximize profit.
problem Statistical flaws and reliance on proxy metrics in A/B tests compromise their effectiveness.
method Hierarchical Bayesian model for estimating conversion probability and monetary value, decision-theoretic stopping rule.
result The framework ensures experiments conclude when no variant offers a significant profit improvement, conserving resources.
Proposes a flexible feature allocation model for sparse factor analysis.
problem Sparse data and rigid assumptions in traditional exploratory tools.
method Adaptive latent feature sharing with control over feature sparsity.
result Derives a novel adaptive Factor analysis (aFA) and aPPCA for flexible dimensionality reduction.
Improved disentangled representation learning using a non-parametric latent density model.
problem Limited disentanglement in VAE due to constraints on latent density independence and complexity.
method Utilized the Indian Buffet Process (IBP) as a non-parametric latent density model to allow richer modeling capacity.
result IBP-VAE outperformed state-of-the-art VAEs in disentangling latent factors across various datasets.