Deep Learning

DOFEN: Deep Oblivious Forest Ensemble achieves state-of-the-art performance on tabular data by using a novel DNN architecture inspired by oblivious decision trees, surpassing other DNNs.

Divide-and-Conquer Posterior Sampling (DCPS) efficiently samples complex posterior distributions from denoising diffusion models (DDMs) for Bayesian inverse problems, significantly improving accuracy …

DISCO: a novel discrete-state continuous-time diffusion model for flexible and efficient graph generation, outperforming state-of-the-art methods.

DISCEDIT efficiently identifies and edits discriminative neural network components for structured pruning and class unlearning, achieving high sparsity and forgetting rates without needing training da…

DIGRAF, a novel graph-adaptive activation function, significantly boosts Graph Neural Network performance by dynamically adapting to graph structure, offering consistent superior results across divers…

DiffusionPDE uses generative diffusion models to solve PDEs accurately, even with highly incomplete observations, outperforming state-of-the-art methods.

Twigs: a novel score-based diffusion framework using multiple co-evolving flows and loop guidance for superior conditional graph generation.

DiffPO: A causal diffusion model learns outcome distributions, offering reliable medical interventions.

Boost classifier robustness with DiffAug, a novel diffusion-based augmentation method! One forward and reverse diffusion step enhances robustness against covariate shifts, adversarial examples, and o…

Derivative-enhanced DeepONets boost PDE solution accuracy and derivative approximation, particularly valuable with limited training data.

DePLM enhances protein optimization by denoising evolutionary information in protein language models via a rank-based diffusion process, improving mutation effect prediction and generalization.

Boost associative memory capacity without extra parameters! DrDAM uses random features to approximate Dense Associative Memories, enabling efficient memory addition and retrieval.

Biologically-inspired Dit-CNNs leverage quadratic neuron integration to capture data correlation, achieving state-of-the-art performance on image classification benchmarks.

DEFT: A novel method efficiently fine-tunes diffusion models for conditional generation via a generalized h-transform, achieving state-of-the-art performance with significant speed improvements.

DeepLag improves fluid prediction by uniquely combining Lagrangian and Eulerian perspectives, tracking key particles to reveal hidden dynamics and improve prediction accuracy.

DeepITE: a novel variational graph autoencoder, efficiently estimates intervention targets from both labeled and unlabeled data, surpassing existing methods in recall and inference speed.

DeepDRK, a novel deep learning approach, significantly improves feature selection by effectively balancing false discovery rate and power, surpassing existing methods, especially with limited data.

A simple, yet accurate model unveils deep learning’s mysteries, providing empirical insights into grokking, double descent, and gradient boosting, offering a new lens for analyzing neural network beha…

PSNR, a novel node-adaptive residual module, significantly improves deep GNN performance by mitigating over-smoothing and handling missing data.

Deep Equilibrium Algorithmic Reasoners (DEARs) achieve superior performance on algorithmic tasks by directly solving for the equilibrium point of a neural network, eliminating the need for iterative r…