Papers on speech emotion recognition, emotional TTS, urban flow and OD demand prediction, and hyperspectral image classification

DUET adds emotion control to frozen diffusion and flow-matching TTS models by steering hidden states along a linearly decodable emotion direction and refining the mel estimate with emotion-recognizer gradients passed through a differentiable vocoder. With GradTTS on ESD it reaches 75.5% average emotion accuracy (strongest supervised baseline: 46.8%).

PhysioSER complements a frozen self-supervised (SSL) backbone with a compact physiology-informed branch that encodes vocal amplitude and phase features with quaternion convolutions, for speech emotion recognition. With frozen WavLM on CREMA-D, it raises weighted accuracy from 69.69% to 75.20%.

MR-UFP predicts the inflow and outflow of city grid regions, even with limited training data, by pre-training encoders with spatial-temporal masking and contrastive learning and training flow prediction jointly with a multi-scale region-classification task. On full TaxiBJ and BikeNYC it beats all 13 baselines (RMSE 14.32 and 4.45).

ST-FCL predicts grid-level urban inflow and outflow by fusing temporal and spatial views, learned through contrastive pretraining, with an external-factor view. On the full TaxiBJ dataset it reaches RMSE 14.71, against 15.41 for the best baseline, ATFM.

MC-STL pre-trains two encoders for urban flow prediction: a ViT encoder learns to reconstruct regions masked at different timestamps, and its attention weights also build a GCN adjacency matrix; a global-local cross-attention encoder learns a temporal-order contrastive task. It reaches RMSE 14.53 on full TaxiBJ.

DSTCN (Dynamic Spatio-Temporal Correlation Network) forecasts origin-destination demand matrices with three modules: Glstm2D for origin- and destination-side demand trends, Simformer for Transformer-based spatial similarity across the OD matrix, and FF-TM for feature fusion and temporal modeling. On NYC-TOD2018 its MAE of 1.469 is 3.04% below the best baseline.

BiST-IF predicts origin–destination (OD) flows between metro stations or urban areas by correcting delayed recent OD matrices, applying bi-directional origin/destination attention, and fusing arrival-side (Out-OD) flows through an attention-based mutual information mechanism. It lowers MAE by an average of 7.55% on HZMetro relative to the best baseline.

S²CMEN (S2CMEN) is a hyperspectral image classification network in which superpixel segmentation and classification enhance each other through a unified loss, combining superpixel-based global spatial context with Spectral-Swin Transformer spectral features. It reaches 97.38% overall accuracy on Indian Pines.

S3CFSL classifies new hyperspectral scenes from five labelled samples per class by transferring knowledge from the labelled Chikusei scene. It combines a cross-spatial–spectral transformer, Gaussian feature denoising and semantic-enhanced domain alignment, and reports 98.52% overall accuracy on Pavia Centre, 88.79% on Salinas and 77.63% on Houston.

A tomato leaf disease detector that adds the DyHead attention module to YOLOv4-tiny and trains it with a Focaler-SIoU box loss. On PlantDoc tomato leaf images it reaches 93.64% mAP, 10.3 percentage points above the YOLOv4-tiny baseline.

URMDet-SimFire is an uncertainty- and reliability-aware multimodal object detection framework for fire and smoke that reweights each input modality by its estimated reliability and calibrates each detection's confidence by its uncertainty. In simulation only, it raised F1 from 0.4348 (vision-only baseline) to 0.4638.