Skip to main content
Search

The University of Tokyo

17 related articles

Subscribe to updates
  1. Fixed transforms no longer need guessing: the Adaptive Interpolating Transform beats fixed-transform baselines with nine added parameters in vision and three per encoder layer in language models

    Fixed transforms such as the discrete Fourier transform and discrete cosine transform are widely used in machine learning to extract features, but choosing among them usually requires knowing the dataset's properties, and the approach weakens when that knowledge is missing. The Adaptive Interpolating Transform (AIT) turns the transform itself into learnable parameters, learning data-driven mappings tailored to the dataset and task. On ImageNet-2012 and the General Language Understanding Evaluation tasks, AIT-enhanced models outperform fixed-transform baselines under both from-scratch and fine-tuning workflows, adding only nine parameters to vision models and three per encoder layer to language models.

  2. arXiv

    A generative sampler's effective sample size collapses with volume in lattice field theory, yet long-distance observables barely move

    Lattice field theory usually judges generative samplers by effective sample size (ESS), a measure of overlap between generated and target samples, but a small global ESS need not mean inaccurate long-distance observables. A scale-resolved overlap diagnostic shows that small errors spread over many short-distance modes can dominate ESS; in two-dimensional φ⁴ theory near criticality, trained diffusion samplers show path-weight ESS fractions falling rapidly with volume while low moments of the magnetization stay close to benchmark values. Given the target marginal, a Metropolis test on a few collective variables keeps an acceptance rate of order one in the tests, and exact local updates then reduce the remaining mismatch.

  3. Nanopore tRNA reads skip basecalling: tyCooNN classifies all 48 E. coli tRNAs at 98.2% accuracy from raw current signals

    When nanopore direct RNA sequencing reads tRNAs, modifications corrupt basecalling, forcing sequence-mapping pipelines to trade classification accuracy against usable reads. tyCooNN (nano tRNA classifier) instead classifies tRNAs directly from ionic current signals, bypassing basecalling and alignment; its ntc48 model reaches 98.2% classification accuracy across all 48 E. coli tRNA species. An equimolar mixture of 44 tRNAs calibrates and corrects read-count bias, yielding absolute abundances, and binary classifiers call specific modifications at the single-molecule level.

  4. arXiv

    ICU sepsis alerts stop at yes-or-no: SepsisFlow puts warning, onset context and a 72-hour course on one hourly clock

    Most ICU sepsis warning models answer only whether sepsis will occur, leaving clinicians unsure of the onset context and early course once an alarm sounds. SepsisFlow places hourly pre-onset risk, the patient state at onset and the 72-hour organ-support trajectory on one clock advancing hourly from ICU admission: before onset it uses only bedside-visible records without antimicrobial or culture orders, reaching a 24-hour AUROC of 0.822 on the MIMIC-IV internal test set, and its onset states form three contexts with 28-day in-hospital mortality of 17.1%, 15.1% and 2.2%.

  5. bioRxiv

    Flowing cells get organelle-level eyes: three structured-light frames plus self-supervised reconstruction reach about 118 nm at up to roughly 720 cells per second

    Super-resolution imaging flow cytometry (SR-IFC) brings structured-illumination microscopy into a continuous microfluidic stream: each cell yields only three phase-shifted frames, physical modulation sharpens the flow axis, and a network trained solely on the imaged sample's own data infers the missing transverse detail. On 100 nm fluorescent beads, SR-IFC reached about 118 nm along the flow axis and 110 nm across it, versus 222 to 227 nm for wide-field imaging of the same beads; with a 2048 × 320 pixel readout, the theoretical throughput ceiling is about 720 cells per second.

  6. arXiv

    Where to inject activation steering: TopoSteer ranks layers by the connectivity length of displacement clouds, skipping held-out trials

    Activation steering adds a vector to a large language model's internal representations at inference time, but where to add it has mostly been settled by trial: existing selectors score how separable the two activation classes are or how well individual displacements align with their mean, without ever steering the model to check. TopoSteer instead scores each layer by the total length of the Euclidean minimum spanning tree of its per-example displacement cloud, measuring whether the mean-difference vector can stand in for those displacements. Across five instruction-tuned models and six binary persona-steering tasks it beats LayerNavigator in most model-budget comparisons, but the margin depends on model and layer budget and is smaller and more mixed on Qwen-7B and Qwen-32B.

  7. Fix the label directions before training: single-cell reference mapping lifts query macro-F1 from 0.774 to 0.811

    Single-cell reference mapping no longer lets class representatives drift during training. SCLSC built each type representative from a mean expression profile, so its position moved as optimization proceeded; SOLAR assigns every reference label a fixed orthogonal direction before training and lets the encoder learn only where cells land in that space. Across nine held-out batches from human immune, lung and pancreas datasets at 30 dimensions, mean query macro-F1 was 0.811, above 0.774 for matched PCA and below 0.838 for scANVI.

  8. arXiv

    A residual-spectrum instability theory explains cascading dimension-wise collapse in VAEs

    The work formulates dimension-wise posterior collapse in variational autoencoders as a fluctuation theory around partially collapsed states: treating the negative ELBO as an effective free energy, its quadratic expansion gives a Hessian acting as a mass matrix, collapsed directions form an invariant fluctuation sector whose exact mass spectrum follows from a conditional residual operator, and a local reactivation direction lowers the free energy once the decoder variance falls below the residual spectral upper edge, with equality marking marginality; numerical continuation shows latent dimensions lost near these spectral marginalities.

  9. arXiv

    Graph neural networks match state-of-the-art performance for positron energy regression and neutron tagging in a Hyper-Kamiokande-like water-Cherenkov detector

    This work applies message-passing graph neural networks to Inverse Beta Decay detection of the Diffuse Supernova Neutrino Background in water-Cherenkov detectors, building a pipeline that reconstructs low-energy events from raw photomultiplier hits: one model regresses 3–30 MeV positron energy and another classifies neutron-capture signals, achieving performance comparable to the state of the art on simulations in a Hyper-Kamiokande-like geometry.

  10. arXiv

    Shared Low-rank Basis Factorization (SLBF) compresses MoE expert weights data-free and beats pruning, merging, and reconstruction baselines across five MoE architectures from 16B to 122B

    The authors analyze three MoE compression families (expert pruning, expert merging, weight reconstruction) and derive structural error bounds showing that pruning and merging incur non-vanishing errors tied to routing and expert heterogeneity, whereas weight reconstruction avoids these structural costs by preserving expert structure and routing; motivated by this, they propose Shared Low-rank Basis Factorization (SLBF), a data-free weight reconstruction method that replaces full-rank shared bases with rank-r bases shared among experts and applies a bilinear gauge fixing to remove redundant parameters, consistently outperforming methods from all three families across five MoE architectures spanning 16B to 122B parameters.