Bayesian SAE model with spectral clustering and uncertainty quantification.
problem Small Area Estimation (SAE) with uncertainty quantification.
method Spectral clustering with external covariates, posterior projections, and CPMSE.
result Closed form expressions for posterior mean estimators and CPMSE.
BatchTopK SAEs improve GPT-2 and Gemma activations with adjustable sparsity.
problem Interpreting language model activations using sparse autoencoders.
method Adapting the TopK constraint to the batch-level, allowing variable number of active latents per sample.
result BatchTopK SAEs consistently outperform TopK SAEs in reconstructing GPT-2 and Gemma activations.
New neural operators learn structured patterns efficiently.
problem Learning and representing complex, structured patterns in data.
method Sparse autoencoder neural operators (SAE-NOs) parameterize concepts as functions, enabling efficient and structured representation.
result SAE-FNOs learn localized patterns and generalize across different scales and discretizations.
AEN-SAEs address feature starvation in sparse autoencoders by stabilizing the geometric alignment of sparse coding.
problem Feature starvation in sparse autoencoders, leading to unstable and misaligned representations.
method Adaptive Elastic Net SAEs (AEN-SAEs) combine ℓ2 and ℓ1 terms to stabilize the sparse coding map and control feature interactions. result AEN-SAEs mitigate feature starvation without heuristic resampling, maintaining competitive reconstruction abilities.
SAE-FiRE extracts key financial info from long documents, improving earnings surprise predictions.
problem Predicting earnings surprises from long, redundant financial documents.
method Sparse Autoencoder feature selection to filter out noise and identify key dimensions.
result SAE-FiRE significantly outperforms baseline approaches in financial datasets.
SAEs struggle with feature consistency across runs, hindering MI reliability.
problem Inconsistency of learned SAE features across different training runs.
method Propose using the Pairwise Dictionary Mean Correlation Coefficient (PW-MCC) to measure feature consistency.
result High levels of feature consistency (0.80 for TopK SAEs on LLM activations) are achievable with appropriate architectural choices.
FF layers in transformers are nearly as interpretable as sparse autoencoders.
problem Comparing interpretability of feature vectors in FF layers vs. sparse autoencoders.
method Revisited interpretability of FF layers as key-value memories using modern benchmarks.
result FF and SAE feature vectors are similarly interpretable, but FFs can be better in some aspects.
New SAE algorithm proves feature recovery for LLMs with theoretical guarantees.
problem Achieving interpretable features in large language models (LLMs).
method Proposed a statistical framework and bias adaptation technique for sparse autoencoders (SAEs).
result Proved correct recovery of all monosemantic features under specific data sampling.
SAE improves VAE's latent space precision in high dimensions.
problem High-dimensional latent codes vs probabilistic inference in VAEs.
method Spherical Auto-Encoder (SAE) with spherical normalization on latent space.
result SAE improves latent code inference precision in high dimensions.
SAEs struggle with curved activation manifolds, revealing layer-dependent scaling laws.
problem Sparse autoencoders' reconstruction error varies across layers, not fitting existing scaling laws.
method Cross-layer study of 844 SAE checkpoints, fitting and regressing on manifold geometry.
result Manifold geometry predicts layer-dependent width exponents in SAEs, with transferable coefficients.
In this paper we propose Structuring AutoEncoders (SAE). SAEs are neural networks which learn a low dimensional representation of data which are additionally enriched with a desired structure in this low dimensional space. While traditional Autoencoders have proven to structure data naturally they fail to discover sema…
Optimal transport offers an alternative to maximum likelihood for learning generative autoencoding models. We show that minimizing the p-Wasserstein distance between the generator and the true data distribution is equivalent to the unconstrained min-min optimization of the p-Wasserstein distance between the encoder agg…
Paper proposes an online speech recognition model using Transformer.
problem Challenges in deploying Transformer-based E2E ASR for online speech recognition.
method Chunk self-attention encoder (chunk-SAE) and monotonic truncated attention (MTA) based self-attention decoder (SAD).
result Achieved 23.66% CER with 320 ms latency, significant improvement over offline models.
Regularization improves stability and consistency of sparse autoencoders.
problem Varying features across random seeds and training choices in SAEs.
method Added L1 or L2 penalties on encoder and decoder weights.
result L2 regularization increases cross-seed feature consistency.
Autoencoders identify brain networks linked to stress and genotype.
problem Designing effective brain stimulation protocols for mental illnesses.
method Supervised autoencoders to model multi-region brain activity.
result Autoencoders reveal a stress-related brain network associated with a bipolar disorder genotype.
New analysis shows interpretability doesn't guarantee steering utility in LLMs.
problem Does higher interpretability lead to better steering utility in large language models?
method Trained 90 SAEs across three LLMs, evaluated interpretability and steering utility, used Kendall's rank coefficients for analysis.
result Interpretability is only weakly associated with steering utility, and features selected by Delta Token Confidence improve steering performance.
The widespread mobile devices facilitated the emergence of many new applications and services. Among them are location-based services (LBS) that provide services based on user's location. Several techniques have been presented to enable LBS even in indoor environments where Global Positioning System (GPS) has low local…
Proposes flexible auto-encoders for varying data dimensions.
problem Fixed latent dimensions limit data flexibility.
method Stochastic bottleneck with weighted dropouts.
result Seamless variable dimensionality reduction with high performance.
Enhanced financial forecasting using supervised autoencoders with noise augmentation and triple labeling.
problem Improving investment strategy performance on noisy financial data.
method Supervised autoencoders with noise augmentation and triple barrier labeling.
result Supervised autoencoders with balanced noise augmentation and bottleneck size significantly boost strategy effectiveness.
We solve a version of the optimal trade execution problem when the mid asset price follows a displaced diffusion. Optimal strategies in the adapted class under various risk criteria, namely value-at-risk, expected shortfall and a new criterion called "squared asset expectation" (SAE), related to a version of the cost v…
Paper hypothesizes MLP layers in LLMs can be approximated by sparse Mixture of Experts.
problem Understanding dense MLP layers in LLMs.
method Theoretical connection between MoE models and SAE structure in activation space.
result MLP layers in LLMs can be well approximated by sparse Mixture of Experts.
One of the most important challenges in the analysis of high-throughput genetic data is the development of efficient computational methods to identify statistically significant Single Nucleotide Polymorphisms (SNPs). Genome-wide association studies (GWAS) use single-locus analysis where each SNP is independently tested…
Deep learning approach for efficient IoT task scheduling in MEC networks.
problem Minimizing task latency in IoT users with large-scale MEC systems.
method Stacked auto-encoder for data compression, adaptive simulated annealing, experience replay.
result Near-optimal performance with significantly reduced computational time.
Deep learning has shown outstanding performance in identifying intricate structures in complex high-dimensional data, especially in the domain of computer vision. The application of deep learning to early detection and automated classification of Alzheimer's disease (AD) has recently gained considerable attention, as r…
Signed-permutation coordinate transport improves model alignment across checkpoints.
problem Improper alignment of coordinate-indexed objects across model checkpoints.
method Introduces sign-marginalized Hungarian matching and coordinate-preserving transport.
result Recovering signed-permutation gauge improves coordinate alignment and model performance.
Unified Bayesian-AI framework improves epidemiological risk prediction and uncertainty quantification.
problem Lack of calibrated uncertainty in machine learning models for epidemiology.
method Combines Bayesian prediction with Bayesian hyperparameter optimization using logistic regression and Gaussian-process Bayesian optimization.
result Unified Bayesian-AI framework provides reliable coverage and improved calibration, enhancing epidemiological decision making.
We exhibit a strong link between frequentist PAC-Bayesian risk bounds and the Bayesian marginal likelihood. That is, for the negative log-likelihood loss function, we show that the minimization of PAC-Bayesian generalization risk bounds maximizes the Bayesian marginal likelihood. This provides an alternative explanatio…
PAC-Bayesian bounds for MLPs with cross entropy loss validated.
problem Generalization bounds for MLPs with cross entropy loss.
method Introduced probabilistic explanations and proved PAC-Bayesian bounds using ELBO.
result MLPs with cross entropy loss inherently guarantee PAC-Bayesian generalization bounds.
Review of priors in Bayesian deep learning models.
problem The importance of prior choices in Bayesian deep learning models.
method Overview of different priors and methods of learning priors from data.
result Motivate practitioners to think carefully about prior specification.
Bayesian methods enhance deep learning models by improving reliability and uncertainty.
problem Improving reliability and uncertainty awareness in deep learning models.
method Approximate Bayesian inference techniques, including SG-MCMC and VI, applied to deep learning models.
result Enhanced posterior inference for deep learning models, particularly in neural networks and generative models.
Bayesian hybrid models correct for missing physics in machine learning.
problem Systematic bias in machine learning models.
method Fusing physics-based insights with machine learning constructs, using Bayesian calibration and stochastic programming.
result Bayesian hybrid models outperform pure machine learning approaches with less data.
Enhances robustness in experimental design through Generalised Bayesian inference.
problem Poor inference and estimates of information gain when statistical model is incorrectly specified.
method Generalised Bayesian (Gibbs) inference framework applied to experimental design.
result GBOED enhances robustness to outliers and incorrect assumptions about noise distribution.
Bayesian neural networks speed up numerical integration.
problem Scalability of Bayesian quadrature methods.
method Bayesian Stein networks using neural networks and Laplace approximation.
result Orders of magnitude speed-up on benchmark functions and real-world problems.
Bayesian uncertainty quantification is flawed, according to new research.
problem Flawed interpretation of Bayesian uncertainty quantification.
method Discussion of Bayesian updating and optimization-based perspective, proposing measures of quality.
result Bayesian uncertainty quantification is not coherent with optimization-based perspective.
Bayesian MAML outperforms MAML in meta learning tasks with theoretical guarantees.
problem Theoretical understanding of Bayesian MAML's superiority over MAML.
method Comparison of meta test risks between Bayesian MAML and MAML in meta linear regression.
result Bayesian MAML has provably lower meta test risks than MAML in both distribution agnostic and linear centroid cases.
One of the main challenges of deep learning tools is their inability to capture model uncertainty. While Bayesian deep learning can be used to tackle the problem, Bayesian neural networks often require more time and computational power to train than deterministic networks. Our work explores whether fully Bayesian netwo…
Bayesian optimization uses shared latent variables for multiple systems.
problem Optimizing systems with limited data and unknown relationships.
method Shared latent variables, Bayesian inference, probabilistic metamodel.
result Performance improvement in zero-, one-, and few-shot settings.
Bayesian REX learns Atari games from demonstrations efficiently.
problem Bayesian reward learning for complex control problems is computationally intractable.
method Bayesian Reward Extrapolation (Bayesian REX) pre-trains a low-dimensional feature encoding and uses preferences to perform fast Bayesian inference.
result Bayesian REX learns Atari games from demonstrations in 5 minutes, competitive with state-of-the-art methods.
Bayesian coresets improve scalable Bayesian inference.
problem Efficiently approximating posterior inference with a subset of data.
method Sparsity constrained optimization and accelerated optimization methods.
result Explicit convergence rate guarantees and superior performance compared to state-of-the-art.
Nonlinear MCMC improves Bayesian machine learning sampling.
problem Sampling problems in Bayesian machine learning.
method Nonlinear MCMC technique with convergence guarantees.
result Improves sampling in Bayesian neural networks.
Discussing hybrid models in Bayesian networks.
problem Improving accuracy in network modeling.
method Hybrid semiparametric Bayesian approach.
result Enhanced model performance in complex networks.
FP-BMA improves generalization by encouraging flat posteriors in Bayesian Model Averaging.
problem Lack of flat posterior in approximate Bayesian inference methods hinders effective Bayesian Model Averaging.
method Proposes Flat Posterior-aware Bayesian Model Averaging (FP-BMA) and Flat Posterior-aware Bayesian Transfer Learning schemes.
result FP-BMA successfully captures flat posteriors, improving generalization performance.
Proposes OBS, a method to adaptively combine Bayesian models online.
problem Learning optimal combinations of Bayesian models in online learning.
method Empirical Bayes lens, Online Bayesian Stacking (OBS).
result Establishes a novel connection between OBS and portfolio selection.
We study the problem of learning Bayesian network structures from data. We develop an algorithm for finding the k-best Bayesian network structures. We propose to compute the posterior probabilities of hypotheses of interest by Bayesian model averaging over the k-best Bayesian networks. We present empirical results on s…
Bayesian neural networks use temperature adjustments to improve predictive performance.
problem Lack of theoretical generalization guarantees for Bayesian neural networks.
method Temperature adjustments to balance likelihood and prior regularization.
result Improved predictive performance through temperature adjustments.
SCoreBO improves Bayesian optimization by learning hyperparameters and self-correcting.
problem Efficient hyperparameter tuning for Gaussian process models in Bayesian optimization.
method Introduces SAL and SCoreBO, which prioritize hyperparameter learning and perform simultaneous optimization and learning.
result SCoreBO outperforms state-of-the-art methods on traditional benchmarks and atypical tasks.
In this paper we introduce ZhuSuan, a python probabilistic programming library for Bayesian deep learning, which conjoins the complimentary advantages of Bayesian methods and deep learning. ZhuSuan is built upon Tensorflow. Unlike existing deep learning libraries, which are mainly designed for deterministic neural netw…
Bayesian optimization with cost-awareness using Gittins index.
problem Optimizing unknown functions with limited data evaluations and costs.
method Developed a connection between cost-aware Bayesian optimization and the Pandora's Box problem, using the Gittins index as an acquisition function.
result The Gittins index-based acquisition function performs well in cost-aware Bayesian optimization, especially in high dimensions.