Agreement versus Softmax under Spatial Disambiguation
[Submitted on 9 Oct 2026] View PDF HTML (experimental) Abstract:ScanRefer Multiple queries require spatial disambiguation: several instances of the target…
Tag: cs.LG
EV-JEPA: Elastic V-JEPA for Any Device and Cross-Device Inference
[Submitted on 9 Oct 2026] View PDF HTML (experimental) Abstract:Foundation models like V-JEPA are increasingly deployed across a heterogeneous edge-cloud…
Next-Concept Modeling: Fast, Parallel Coarse-to-Fine Text Generation
[Submitted on 9 Oct 2026] View PDF HTML (experimental) Abstract:Diffusion language models (DLMs) offer parallel sampling for faster generation than…
Human Pose Estimation by Probabilistic Mixtures of Gaussian and Uniform Distributions
[Submitted on 8 Oct 2026] View PDF HTML (experimental) Abstract:Despite advancements in sub-pixel decoding, modern Human Pose Estimation (HPE) architectures…
Learning $omega$-Powered Conditional Distributions with Importance-Weighted Flow Matching
[Submitted on 12 Oct 2026] View PDF HTML (experimental) Abstract:Classifier-Free Guidance (CFG) with guidance scale $gamma$ is commonly motivated by…
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models
[Submitted on 9 Oct 2026] View PDF Abstract:Generalist robot policies are trained with imitation learning and are bounded by the…
STAPLE: Static Sparse Upcycling of Pretrained Checkpoints via Embedding Modules
[Submitted on 8 Oct 2026] Authors:Ranajoy Sadhukhan, Euijin Hong, Sheng Cao, Zechun Liu, Shuowei Jin, Sha Li, Changlong Yu, Xin…
Learning by Compression: Risk Estimates Attain the Law-of-Large-Numbers Rate
[Submitted on 9 Oct 2026] View PDF HTML (experimental) Abstract:Compression schemes provide a general framework for deriving data-dependent risk bounds…
Efficient Learning of Survival Analysis Models Under Outcome-Dependent Truncation and Censoring
[Submitted on 8 Oct 2026] View PDF HTML (experimental) Abstract:Modern machine learning models are typically trained on data assumed to…
How Well Do Prior-Data Fitted Networks Amortize Bayesian Inference?
[Submitted on 9 Oct 2026] View PDF HTML (experimental) Abstract:Prior-data fitted networks (PFNs) are trained to approximate the Bayesian posterior…