For uniform keys on the d-dimensional sphere, softmax attention becomes selective at inverse temperature scaling β_n* ≍ n^{2/(d-1)}, with explicit limiting laws for attention weights and outputs in each regime.
hub
Proceedings of the IEEE/CVF international conference on computer vision , pages=
27 Pith papers cite this work. Polarity classification is still indexing.
27
Pith papers citing it
hub tools
citation-role summary
background 2
method 1
citation-polarity summary
representative citing papers
A neurosymbolic model augments Swin Transformers with focal sets and fuzzy logic to produce calibrated hierarchical image classifications that respect logical constraints.
Polyphonia improves zero-shot stem-specific timbre transfer in polyphonic music by 15.5% target alignment via acoustic-informed attention calibration that uses probabilistic priors to set coarse boundaries.
LE-SAM inverts SAM by fixing the loss budget instead of the parameter-space radius, yielding better generalization across benchmarks.
ALiBi bias is the expectation of positional LSH-induced block masks, yielding spectral and max-norm approximation bounds that reduce long-context biased attention to randomized short-context unbiased attention.
Empirical tests with quad-mesh filling indicate that decision regions in modern image classifiers are simply connected.
Adversarial training on simplified Vision Transformers achieves benign overfitting with near-zero robust loss and generalization error when signal-to-noise ratio and perturbation budget meet specific conditions.
NodePFN pre-trains on synthetic graphs with controllable homophily and causal feature-label models to achieve 71.27 average accuracy on 23 node classification benchmarks without graph-specific training.
DGNO parameterizes integral kernels with discontinuous Galerkin elements for heterogeneous defocus deblurring in pathology images and reports superior performance over prior methods.
ST-TGExplainer disentangles stability and transition patterns in temporal graphs via a self-explainable TGNN guided by a disentangled information bottleneck objective to produce more faithful explanations.
ElasticDiT introduces an elastic DiT architecture with adjustable spatial compression and block depth plus Shift Sparse Block Attention and a distilled VAE to enable a single model to cover multiple fidelity-latency points for high-resolution image generation on mobile devices.
ROMER cuts perplexity by up to 59% in noisy analog CIM environments for MoE LLMs via expert replacement and router recalibration calibrated on real-chip measurements.
A frequency-enhanced Vision Transformer with FDSA, FGMLP, WAFF, and FCSB modules delivers superior volumetric medical image segmentation performance and efficiency over prior state-of-the-art methods.
ReSIDe generalizes logit-based confidence scores to intermediate layers of synthetic image detectors and uses preference optimization to aggregate them, cutting area under the risk-coverage curve by up to 69.55% under covariate shifts.
In LVLMs, attention can be replaced by random Gaussian weights with little or no performance loss, indicating that current models get lost in attention rather than efficiently using visual context.
Hallucinations in diffusion models are driven by local intrinsic dimension instabilities on the manifold, which Intrinsic Quenching corrects by deflating it.
HEXST applies a hexagonal shifted-window Transformer with rotary positional encodings, contrast-sensitive training objectives, and single-cell priors to predict gene expression from histology slides, outperforming prior models on seven datasets while preserving spatial heterogeneity.
SalUn uses gradient-based weight saliency to achieve effective machine unlearning of data, classes, or concepts in image classification and generation, narrowing the gap to exact retraining.
PACD-Net uses pseudo-augmented contrastive distillation with a hybrid Swin Transformer-CNN backbone to estimate TAR, TIR, and TBR from sparse SMBG data and outperforms prior methods in accuracy and stability under sparse conditions.
STAR-IOD applies scale-decoupled topology alignment and K-Means-based pseudo-label refinement to reduce catastrophic forgetting in remote sensing incremental object detection, reporting 1.7% and 2.1% mAP gains on new DIOR-IOD and DOTA-IOD datasets.
Semi-LAR is a semi-supervised contrastive learning framework with linear attention for nighttime flare removal that refines pseudo-labels via quality assessment and uses flare-aware patch-level contrastive losses.
TAPE applies temporal-aware token pruning with smoothing, reselection, and timestep scheduling to speed up video diffusion models while preserving visual fidelity and coherence.
Active learning with randomly initialized models achieves comparable results to traditional candidate-model methods, with low-confidence sampling proving most effective.
TaTok is a theoretically grounded adaptive tokenization method that uses global tokens and cumulative conditional entropy filtering to reduce redundancy while improving reconstruction quality over fixed-rate patch tokenization.
citing papers explorer
-
ROMER: Expert Replacement and Router Calibration for Robust MoE LLMs on Analog Compute-in-Memory Systems
ROMER cuts perplexity by up to 59% in noisy analog CIM environments for MoE LLMs via expert replacement and router recalibration calibrated on real-chip measurements.
-
SalUn: Empowering Machine Unlearning via Gradient-based Weight Saliency in Both Image Classification and Generation
SalUn uses gradient-based weight saliency to achieve effective machine unlearning of data, classes, or concepts in image classification and generation, narrowing the gap to exact retraining.