Skip to main content
Search

Singapore University of Technology and Design

5 related articles

Subscribe to updates
  1. arXiv

    One Model, Whole Energy Curve: Nuclei Join the Network So N2 Dissociation Needs No Retraining

    NucleoNet evolves nuclei and electrons together in one network, so a single parameter set describes the N2 dissociation curve, Li displacement in LiCN, and the ethane and formamide bond-cleavage paths, matching established neural-wavefunction accuracy on equilibrium small molecules with far fewer parameters.

  2. arXiv

    WiSPER combines pose-supervised pretraining with residual flow refinement to cut WiFi-CSI multi-person 3D pose error to 63.72 mm MPJPE

    WiSPER is a two-stage framework: PAMEL couples JEPA-style masked latent prediction with auxiliary pose-set supervision on the same visible CSI context to guide the encoder toward joint localization from partial observations, and ReFT refines pose candidates through a conditional residual flow guided by coarse coordinates and per-joint decoder features. On PiW3D it reaches 63.72 mm overall MPJPE, a 40.0% reduction relative to WiFi-JEPA, with 42.1% and 38.1% reductions for two- and three-person observations.

  3. arXiv

    Recognizing Signomial Convexity Is Shown Strongly NP-Hard, Closing All Three Tractability Routes

    The work studies convexity recognition for signomials (finite sums of generalized monomials with real exponents) and shows that deciding whether a signomial is convex in its original variables, becomes convex after a logarithmic change of variables, or, when positive, becomes convex after additionally taking the logarithm of its value (the structure underlying disciplined geometric programming) is strongly NP-hard, both on compact domains and globally, via reductions from gap-promise variants of polynomial convexity that preserve curvature gaps under the relevant logarithmic transformations.

  4. arXiv

    DC-SAE splits image tokenization into semantic and pixel branches, hitting 32x compression with 3.37 gFID and 29.79 PSNR on ImageNet and a 4.41x speedup over DC-AE

    Peking University and Singapore University of Technology and Design propose DC-SAE, a decoupled compact semantic autoencoder that pairs a frozen semantic encoder for a generation-friendly compact latent with a trainable pixel encoder for low-level detail, achieving 29.79 PSNR and 3.37 gFID on ImageNet at 32x spatial compression, improving over DC-AE by 13.5% in PSNR and 54.9% in gFID with a 4.41x speedup, while a B-parameter DiT reaches 0.84 on GenEval and 86.007 on DPG-Bench.

  5. arXiv

    BlowLive combines blow-acoustic and facial biometrics into a multi-factor authenticator, reaching 100% accuracy on fused modalities from 50 participants with template and key revocation

    The work proposes and implements BlowLive, a multi-factor biometric authentication framework that treats the acoustic signal of blowing onto a phone as a behavioral modality and a simultaneously captured face as a physiological modality, generating blow embeddings via GFCC features plus a CNN and facial embeddings via FaceNet, then deriving cryptographic keys from binarized fused embeddings through a fuzzy extractor, with an added Doppler-shift liveness detection module; on data collected from 50 participants it reports 99.56% accuracy for blow-acoustics and 100% for facial and fusion modalities, and 99.46% liveness detection accuracy under per-user thresholds.

You’re all caught up