Deep Learning

RandNet-Parareal: A novel time-parallel PDE solver using Random Neural Networks achieves speed gains up to x125, dramatically improving scalability for large-scale simulations.

Quantum VAE with spherical latent variable learning enables efficient, one-shot 3D molecule generation, outperforming classic and other quantum methods.

Quasi-Bayesian Vine (QB-Vine) efficiently models high-dimensional densities by recursively updating 1D marginal predictives and a vine copula, significantly outperforming state-of-the-art methods.

Quantum Deep Equilibrium Models (QDEQs) achieve higher QML performance with shallower circuits by using a DEQ training paradigm, improving near-term quantum computation efficiency.

PUREGEN uses generative model dynamics to purify poisoned training data, providing a universal, effective, and efficient train-time defense against various data poisoning attacks.

PURE: A novel method uses Graph ODE to adapt spatio-temporal forecasting models to various fluid dynamics scenarios, improving model adaptation to unseen parameters and long-term predictions.

DASH: a novel pruning framework leverages domain knowledge to improve the interpretability and sparsity of neural network models for gene regulatory dynamics, outperforming existing methods.

Provably robust diffusion posterior sampling for plug-and-play image reconstruction is achieved via a novel algorithmic framework, DPnP, offering both asymptotic and non-asymptotic performance guarant…

Boosting posterior sampling in challenging high-dimensional inverse problems, this paper introduces ’tilted transport’, a novel technique leveraging denoising oracles for provably easier sampling.

One data point per class suffices for efficient and provable dataset distillation in kernel ridge regression, significantly reducing computational costs.

ProSST, a novel protein language model, integrates protein sequences and structures using quantized structure representation and disentangled attention, achieving state-of-the-art performance in zero-…

This study introduces a novel Bayesian functional mixed model that reliably recovers the size and shape of fixed effects from noisy functional data with phase variability, outperforming current state-…

IPR-MPNNs revolutionize graph neural networks by implicitly rewiring graphs using virtual nodes, achieving state-of-the-art performance with significantly faster computation.

Probabilistic Decomposed Linear Dynamical Systems (p-dLDS) improve latent variable inference in nonlinear neural systems by using a probabilistic approach that’s robust to noise and includes a time-va…

Random noise pretraining dramatically speeds up and enhances neural network learning without weight transport, mimicking the brain’s developmental process and achieving performance comparable to backp…

Eliciting high-dimensional probability distributions from experts using only preference comparisons is achieved via normalizing flows and a novel functional prior, resolving the problem of collapsing …

Predictive Attractor Models (PAM) offer a biologically-plausible, streaming sequence memory architecture that avoids catastrophic forgetting and generates multiple future possibilities.

Deep learning algorithms now predict quantum ground state properties with constant sample complexity, regardless of system size, improving upon previous methods.

Researchers achieved high-accuracy differentially private (DP) models by using a novel DP continual pre-training strategy with only 10% public data, mitigating the performance degradation common in DP…

Revolutionizing unordered data compression, this paper introduces autoregressive shuffle coding, achieving state-of-the-art speeds and compression rates on massive datasets.