GeoMix achieves new state-of-the-art results in descriptor-free 2D-3D matching by adding directional embeddings, learnable global context nodes, and multi-detector training, cutting rotation and translation errors by up to 90% on standard benchmarks.
hub
International conference on machine learning , pages=
17 Pith papers cite this work. Polarity classification is still indexing.
17
Pith papers citing it
hub tools
citation-role summary
background 1
method 1
citation-polarity summary
representative citing papers
PatternGSL introduces a learnable specification language for sewing patterns that lets vision-language models reconstruct explicit, simulation-ready 3D garments from single images, backed by a new 300K paired dataset.
The paper proves negative weight drift at initialization under MSE or cross-entropy with asymmetric activations, links it to up to 90% sparsity in GPT-nano, maps the sparsity-accuracy cliff across 79 configurations, and shows clipped ReLU² and GELU² improve validation loss.
A fully differentiable TensorFlow gyrokinetic code allows approximate gradients of nonlinear turbulence quantities to be used for outer-loop tasks such as profile prediction despite stochasticity.
A CNN trained on AREPO simulations and synthetic observations reverts edge-on 13CO spectral data to top-down views of the CMZ as a proof-of-concept for supervised reversion.
MixTTA equips normalization layers with low-rank cross-channel transformations plus decoupling and spectral projections to correct cross-channel structural changes during test-time adaptation.
PHALAR achieves up to 70% relative accuracy gain in stem retrieval over prior art using under half the parameters and 7x faster training by enforcing musical equivariances via spectral pooling and complex heads.
SPIN lets weak LLMs become strong by self-generating training data from previous model versions and training to prefer human-annotated responses over its own outputs, outperforming DPO even with extra GPT-4 data on benchmarks.
ST-MoE introduces stability techniques for sparse expert models, allowing a 269B-parameter model to achieve state-of-the-art transfer learning results across reasoning, summarization, and QA tasks at the compute cost of a 32B dense model.
MATE uses permutation-invariant sum-aggregated memory of transition embeddings to solve CMDPs with online adaptation and computational advantages over Transformers and RNNs.
Learning a margin-based confidence ranker for LLM judges improves agreement-target success in cascaded selective evaluation compared to heuristic confidence scores.
SSL clustering is derived as KL-divergence optimization where a teacher-distribution constraint normalizes via inverse cluster priors and simplifies to batch centering by Jensen's inequality.
A pre-activation regularizer seeds more affine regions near data in piecewise affine networks, increasing local region count and improving early training performance.
Covariance-aware ridge and combined l1-l2 regularizers for neural networks yield better predictive performance and complexity control than standard penalties in simulations and applications to cooling-load prediction and leukemia classification.
Neo, a cGAN, super-resolves HSC images to HST-like quality and improves galaxy morphological parameter accuracy by factors of 2-10.
Linkify augments assembly graphs with corrected interface point clouds and trains GATv2 for masked part prediction, outperforming non-graph baselines on Fusion 360 data.
SNAC-Pack uses learned FPGA resource surrogates inside evolutionary neural architecture search to produce smaller, faster FPGA models for jet classification and qubit readout than BOP-based codesign.
citing papers explorer
-
GeoMix: Descriptor-Free Visual Localization via Global Context and Multi-Detector Training
GeoMix achieves new state-of-the-art results in descriptor-free 2D-3D matching by adding directional embeddings, learnable global context nodes, and multi-detector training, cutting rotation and translation errors by up to 90% on standard benchmarks.
-
PatternGSL: A Structured Specification Language for Template-Free and Simulation-Ready 3D Garments
PatternGSL introduces a learnable specification language for sewing patterns that lets vision-language models reconstruct explicit, simulation-ready 3D garments from single images, backed by a new 300K paired dataset.
-
Bug or Feature$^2$: Weight Drift, Activation Sparsity and Spikes
The paper proves negative weight drift at initialization under MSE or cross-entropy with asymmetric activations, links it to up to 90% sparsity in GPT-nano, maps the sparsity-accuracy cliff across 79 configurations, and shows clipped ReLU² and GELU² improve validation loss.
-
iGENE: A Differentiable Flux-Tube Gyrokinetic Code in TensorFlow
A fully differentiable TensorFlow gyrokinetic code allows approximate gradients of nonlinear turbulence quantities to be used for outer-loop tasks such as profile prediction despite stochasticity.
-
IRIS: Deciphering Spectral-Line Imagery of the Galactic Center by Machine-Learning on Simulations
A CNN trained on AREPO simulations and synthetic observations reverts edge-on 13CO spectral data to top-down views of the CMZ as a proof-of-concept for supervised reversion.
-
MixTTA: Low-Rank Cross-Channel Mixing for Reliable Test-Time Adaptation
MixTTA equips normalization layers with low-rank cross-channel transformations plus decoupling and spectral projections to correct cross-channel structural changes during test-time adaptation.
-
PHALAR: Phasors for Learned Musical Audio Representations
PHALAR achieves up to 70% relative accuracy gain in stem retrieval over prior art using under half the parameters and 7x faster training by enforcing musical equivariances via spectral pooling and complex heads.
-
Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models
SPIN lets weak LLMs become strong by self-generating training data from previous model versions and training to prefer human-annotated responses over its own outputs, outperforming DPO even with extra GPT-4 data on benchmarks.
-
ST-MoE: Designing Stable and Transferable Sparse Expert Models
ST-MoE introduces stability techniques for sparse expert models, allowing a 269B-parameter model to achieve state-of-the-art transfer learning results across reasoning, summarization, and QA tasks at the compute cost of a 32B dense model.
-
MATE: Solving Contextual Markov Decision Processes with Memory of Accumulated Transition Embeddings
MATE uses permutation-invariant sum-aggregated memory of transition embeddings to solve CMDPs with online adaptation and computational advantages over Transformers and RNNs.
-
Margin-Adaptive Confidence Ranking for Reliable LLM Judgement
Learning a margin-based confidence ranker for LLM judges improves agreement-target success in cascaded selective evaluation compared to heuristic confidence scores.
-
Information theoretic underpinning of self-supervised learning by clustering
SSL clustering is derived as KL-divergence optimization where a teacher-distribution constraint normalizes via inverse cluster priors and simplifies to batch centering by Jensen's inequality.
-
Region Seeding via Pre-Activation Regularization: A Geometric View of Piecewise Affine Neural Networks
A pre-activation regularizer seeds more affine regions near data in piecewise affine networks, increasing local region count and improving early training performance.
-
Adaptive Norm-Based Regularization for Neural Networks
Covariance-aware ridge and combined l1-l2 regularizers for neural networks yield better predictive performance and complexity control than standard penalties in simulations and applications to cooling-load prediction and leukemia classification.
-
Photometric Super-Resolution for Improving Galaxy Morphological Measurements using Conditional Generative Adversarial Networks
Neo, a cGAN, super-resolves HSC images to HST-like quality and improves galaxy morphological parameter accuracy by factors of 2-10.
-
Linkify: Learning from Interface-Augmented Assembly Graphs
Linkify augments assembly graphs with corrected interface point clouds and trains GATv2 for masked part prediction, outperforming non-graph baselines on Fusion 360 data.
-
SNAC-Pack 2.0: Scaled-Out Surrogate Neural Architecture Codesign
SNAC-Pack uses learned FPGA resource surrogates inside evolutionary neural architecture search to produce smaller, faster FPGA models for jet classification and qubit readout than BOP-based codesign.