Skip to main content
Search

# Domain Adaptation

15 related articles

Subscribe to updates
  1. arXiv

    Ancient-terrace point clouds fail across regions: MELD corrects each feature level with unlabeled target data, lifting F1 from 0.397 to 0.459

    Between the Teyuna and Miraflores lidar surveys in the Sierra Nevada de Santa Marta, Colombia, terrace detectors trained in one region degrade in the other because terrain, vegetation, acquisition settings and terrace morphology differ. MELD trains a separate diffusion denoiser on unlabeled target features at each of four PointNet++ levels, then uses these priors to refine the multi-scale features fed to the segmentation decoder during training and inference. F1 rises from 0.397 to 0.459 for Teyuna-to-Miraflores transfer and from 0.557 to 0.594 in reverse, the highest among the compared adaptation methods in both directions; adding KNN local-geometry descriptors raises these to 0.542 and 0.606.

  2. medRxiv

    Fetal ultrasound screening learns from scarce labels: RadarMAE's radar-style masking holds cross-hospital metric drops to 0.04

    First-trimester ultrasound images for Down syndrome screening are low-resolution and structurally variable, and confirmed rare cases are scarce, so standard self-supervised pretraining struggles. RadarMAE instead masks alternating concentric rings and angular sectors, a radar-like pattern that forces the network to learn the continuous curvature and elongated shapes of craniofacial structures, reaching AUC 0.99 and 97% accuracy on an internal cohort. With domain-adversarial adaptation, the model calibrates on unlabeled images from four external hospitals and limits the drop in key metrics to 0.04 or less.

  3. bioRxiv

    Cell-line drug sensitivity must reach patients: MODAE+ aligns the two domains while keeping tumour lineage, and its predicted sensitivity tracks survival

    Transferring drug sensitivity from cancer cell lines to patient tumours is hard because aligning the two domains can erase tumour-lineage structure. MODAE+ jointly optimizes domain alignment, deconfounding, drug-response prediction, survival prediction and lineage classification in one shared latent space, with adaptive multitask weighting to balance them. On matched treated TCGA subsets, its predicted sensitivity reached a 41.5% Cox-significant rate for overall survival, versus 4.9% for CODE-AE-ADV.

  4. Nature Methods

    DIRAC folds spatial multi-omic integration and annotation into one model, and finds a microvascular proliferation domain other methods missed in glioblastoma

    Spatial multi-omic data pose two jobs at once: merging molecular layers such as RNA and protein measured on the same tissue section, and transferring cell-type labels from a reference dataset onto a new sample. DIRAC treats both as domain adaptation, using a graph neural network and an adversarial discriminator to learn embeddings that are insensitive to batch and modality but discriminative for cell type. On simulated and real DBiT-seq data it beat comparison methods on domain recovery and mixing, and in glioblastoma tissue it identified a microvascular proliferation domain the other methods missed.

  5. Rat sleep staging holds up when the drug changes: trained on saline alone, F1 falls less than 1% under psilocybin and MDL100907

    On more than 1.1 million annotated rat EEG/EMG segments, a robustness-oriented deep learning pipeline reaches 94.3% F1 for four-class sleep staging, 7.3 and 6.4 points above SlumberNet's 87.0% and random forest's 87.9%. Most automated stagers were evaluated on random splits that preserve the training distribution, and degrade when subjects or drug conditions shift. The pipeline combines a global context vector, supervised contrastive learning and a dual-branch classifier to learn representations insensitive to subject and treatment; trained on saline recordings only and tested on unseen animals, every metric drops by less than 1%.

  6. arXiv

    Multi-domain fine-tuning stops stepping on itself: 256 examples per domain lift expert routing to 90–98%

    Dense Transformer layers are expanded into sparse mixture-of-experts layers, giving law and biomedicine their own full MLP experts while a router is supervised with a small number of examples to pick the right expert per domain. On Llama-3.1-8B and Qwen-3-8B the method posts the highest biomedical and legal averages and the highest general-domain average; with only 256 examples per domain for router supervision, routing accuracy on specialized domains reaches 90–98% after adaptation, against 43–70% for routers trained without it.

  7. Parkinson's EEG microstates fail to travel: cohort origin is decodable at AUC 0.966 in the same feature space

    Expanding resting-state EEG A/B/C/D microstate sequences into static, directed-transition, flow and complexity features separated Parkinson's disease from controls within a development cohort at ROC-AUC 0.823, but locked testing on an independent OpenNeuro cohort fell to 0.513, near chance. In that same feature space, a classifier predicting dataset origin reached ROC-AUC 0.966, showing the two cohorts' feature distributions differ sharply. The study therefore evaluates within-cohort discrimination and cross-cohort transportability separately, and argues for explicit distribution-shift checks before microstate features are treated as reproducible electrophysiological markers.

  8. Metabolomics

    A heart-failure metabolite model breaks on a second NMR platform: 108 duplicate samples lift balanced accuracy from 0.64 to 0.72

    Metabolomics models trained on one cohort often lose accuracy on another because platform, sample handling and quantification differences shift the data. Using 12,000 plasma NMR samples from the Estonian Biobank to classify heart failure (ICD-10 I50.9), a study evaluated 295 samples re-measured on a different NMR platform and calibrated alignment with up to 108 biospecimens duplicated across both. Harmonization generally beat unaligned data and stabilized as matched pairs grew, yet simple within-domain normalization stayed a strong baseline, reaching balanced accuracy 0.72 and MCC 0.46 with a support vector machine at all 108 pairs.

  9. arXiv

    OnePO Adapts Base Models to Medicine with RL Alone, as HuatuoGPT-3-27B Reaches 71.4 on HealthBench Professional

    The work proposes One-stage Policy Optimization (OnePO), which treats teacher outputs as transient guidance rather than persistent training targets: a probability floor and gradient rescaling strengthen learning on low-probability teacher tokens, and Teacher Retirement discards a teacher output once the current policy's reward surpasses it. With only 20K medical samples, OnePO reaches 67.2 on HealthBench (Total), outperforming SFT+RL and pure RL under the same teacher, and scales to the open-source HuatuoGPT-3 series, whose 27B variant reaches 71.4 on HealthBench Professional.

  10. arXiv

    AutoAdapt trains independent LoRA adapters via automatic latent domain discovery, matching an all-domain single adapter across 14 benchmarks while enabling retraining-free extension

    AutoAdapt is a modular instruction-tuning framework that first discovers latent domains with three methods (K-means, BERTopic, MNLI), then trains one LoRA adapter per domain independently and in parallel, and routes at inference with a parameter-free centroid mechanism; across 14 domain-specific benchmarks and GPT-4o pairwise judgements it reaches parity with a single LoRA adapter trained on all domains, while adding or updating a domain requires training only one adapter instead of full-model retraining.