# Mamba

Deep learning models for electrocardiograms are mostly trained with supervision tied to specific diagnostic labels, which makes them fragile when patient populations or acquisition settings change. This self-supervised pretraining framework learns features from 918,915 unlabeled 12-lead ECG recordings, with one path explicitly measuring each lead's deviation from normal waveform morphology and another reconstructing cross-lead associations under those deviation cues. With the feature extractor frozen and no downstream fine-tuning, it improves AUROC by up to 6.5% and F1 by up to 12.8% over prior methods on four public datasets.

Deep learning models for electrocardiograms are mostly trained with supervision tied to specific diagnostic labels, which makes them fragile when patient populations or acquisition settings change. This self-supervised pretraining framework learns features from 918,915 unlabeled 12-lead ECG recordings, with one path explicitly measuring each lead's deviation from normal waveform morphology and another reconstructing cross-lead associations under those deviation cues. With the feature extractor frozen and no downstream fine-tuning, it improves AUROC by up to 6.5% and F1 by up to 12.8% over prior methods on four public datasets.

Arctic sea ice concentration forecasting has been limited by the difficulty of fusing multi-timescale information, with statistical models missing nonlinear relationships and numerical models depending on initial conditions. STICEnet replaces the U-Net encoder-decoder path with strictly unidirectional temporal causal and transposed convolutions and embeds channel and spatial attention in the skip connections, trained on NSIDC sea ice concentration and ERA5 atmospheric and oceanic variables. Across one- to six-month leads its RMSE, MAE and R2 all beat CNN, ConvLSTM, IceNet and MT-IceNet, reaching 7.51% RMSE and 0.925 R2 at month six.

Arctic sea ice concentration forecasting has been limited by the difficulty of fusing multi-timescale information, with statistical models missing nonlinear relationships and numerical models depending on initial conditions. STICEnet replaces the U-Net encoder-decoder path with strictly unidirectional temporal causal and transposed convolutions and embeds channel and spatial attention in the skip connections, trained on NSIDC sea ice concentration and ERA5 atmospheric and oceanic variables. Across one- to six-month leads its RMSE, MAE and R2 all beat CNN, ConvLSTM, IceNet and MT-IceNet, reaching 7.51% RMSE and 0.925 R2 at month six.

NVIDIA's Nemotron 3.5 Lightning is now free on AI Gateway as nvidia/nemotron-3.5-lightning-free, a 30B-parameter model with 3B active per token aimed at long-running agents and RAG systems.

NVIDIA's Nemotron 3.5 Lightning is now free on AI Gateway as nvidia/nemotron-3.5-lightning-free, a 30B-parameter model with 3B active per token aimed at long-running agents and RAG systems.

The work proposes TranSamba, which places a Cross-Plane Mamba (CPM) block before the Transformer block in every layer of a DeiT-S Vision Transformer encoder so that patch tokens in neighboring planes exchange information in linear time, improving in-plane self-attention and the class-to-patch attention maps; trained with only plane-level labels, 16 contiguous planes per volume and 16 volumes per batch for 100 epochs, it reaches 64.1%, 26.8% and 41.0% DSC and 46.8%, 25.8% and 25.6% IoU on the BraTS, KiTS and LASC test sets, the highest among the compared methods, with time complexity linear in the number of planes and batch space complexity independent of it.

The work proposes TranSamba, which places a Cross-Plane Mamba (CPM) block before the Transformer block in every layer of a DeiT-S Vision Transformer encoder so that patch tokens in neighboring planes exchange information in linear time, improving in-plane self-attention and the class-to-patch attention maps; trained with only plane-level labels, 16 contiguous planes per volume and 16 volumes per batch for 100 epochs, it reaches 64.1%, 26.8% and 41.0% DSC and 46.8%, 25.8% and 25.6% IoU on the BraTS, KiTS and LASC test sets, the highest among the compared methods, with time complexity linear in the number of planes and batch space complexity independent of it.