Just out: Al-doped C3N4 composite cathodes suppress polysulfide shuttle and enable long-lifetime Li-S batteries with physics-informed NeuralODE lifetime forecasting
Al-doped C3N4 composite cathodes suppress polysulfide shuttle and enable long-lifetime Li-S batteries with physics-informed NeuralODE lifetime forecasting
Communications Chemistry, Published online: 02 October 2026; doi:10.1038/s42004-026-02220-2Lithium-sulfur (Li-S) batteries offer substantial gains in theoretical energy density but are hampered in practice by polysulfide shuttling, poor electronic conductivity and complex, stage-dependent degradation. Here, the authors synthesize and systematically evaluate seven sulfur-composite cathodes and identify an Al-doped graphitic carbon nitride (Al-C3N4/S) composite that reduces sheet resistance, strengthens electrode microstructure and enhances polysulfide immobilization, interpreting and forecasting degradation through a hybrid NeuralODE framework that integrates physics-based data augmentation with continuous-time neural dynamics.
dlvr.it
October 2, 2026 at 1:53 AM
Just out: Al-doped C3N4 composite cathodes suppress polysulfide shuttle and enable long-lifetime Li-S batteries with physics-informed NeuralODE lifetime forecasting
In #Microbiome
Deciphering microbial community dynamics using cross-sectional data-informed NeuralODE🦠
Read more ➡️ doi.org/rm9x #Microbes
Deciphering microbial community dynamics using cross-sectional data-informed NeuralODE🦠
Read more ➡️ doi.org/rm9x #Microbes
September 9, 2026 at 9:08 AM
In #Microbiome
Deciphering microbial community dynamics using cross-sectional data-informed NeuralODE🦠
Read more ➡️ doi.org/rm9x #Microbes
Deciphering microbial community dynamics using cross-sectional data-informed NeuralODE🦠
Read more ➡️ doi.org/rm9x #Microbes
SUDO solves unbalanced optimal transport for any convex growth penalty—no expensive NeuralODE simulation required. Matches analytical WFR solutions while supporting asymmetric proliferation priors on real single-cell data. …
https://bymachine.news/sudo-unbalanced-optimal-transport-cellular-dynamics
https://bymachine.news/sudo-unbalanced-optimal-transport-cellular-dynamics
SUDO Framework Enables Efficient Inference of Cellular Growth Dynamics
SUDO framework enables simulation-free inference of cellular dynamics with general growth penalties, matching analytical solutions without NeuralODE overhead.
bymachine.news
September 7, 2026 at 5:30 PM
SUDO solves unbalanced optimal transport for any convex growth penalty—no expensive NeuralODE simulation required. Matches analytical WFR solutions while supporting asymmetric proliferation priors on real single-cell data. …
https://bymachine.news/sudo-unbalanced-optimal-transport-cellular-dynamics
https://bymachine.news/sudo-unbalanced-optimal-transport-cellular-dynamics
Worth noting that even in the low-dim periodic VDP, the LC loss does seem to help a little bit (the NeuralODE that also sees noisy data doesn't do quite as well). So it seems to be helpful to enforce VF conservativity along directions orthogonal to the flow
November 28, 2025 at 5:05 AM
Worth noting that even in the low-dim periodic VDP, the LC loss does seem to help a little bit (the NeuralODE that also sees noisy data doesn't do quite as well). So it seems to be helpful to enforce VF conservativity along directions orthogonal to the flow
- Heterogeneous node and edge dynamics
- Modular equation-based modeling
- Symbolic simplification
- Parallel on CPU and GPU (experimental)
- Automatic Differentiation compatible (enables NeuralODE on nodes, parameter fitting, sensitivity analysis, optimization of dynamic properties, etc...)
- Modular equation-based modeling
- Symbolic simplification
- Parallel on CPU and GPU (experimental)
- Automatic Differentiation compatible (enables NeuralODE on nodes, parameter fitting, sensitivity analysis, optimization of dynamic properties, etc...)
October 15, 2025 at 8:27 AM
- Heterogeneous node and edge dynamics
- Modular equation-based modeling
- Symbolic simplification
- Parallel on CPU and GPU (experimental)
- Automatic Differentiation compatible (enables NeuralODE on nodes, parameter fitting, sensitivity analysis, optimization of dynamic properties, etc...)
- Modular equation-based modeling
- Symbolic simplification
- Parallel on CPU and GPU (experimental)
- Automatic Differentiation compatible (enables NeuralODE on nodes, parameter fitting, sensitivity analysis, optimization of dynamic properties, etc...)
A new study shows that guiding Neural ODEs with a low‑dimensional manifold improves accuracy, cuts function evaluations and speeds training convergence on benchmark datasets. Read more: https://getnews.me/manifold-constrained-neural-ode-improves-speed-and-accuracy/ #neuralode #manifoldlearning
October 7, 2025 at 10:59 PM
A new study shows that guiding Neural ODEs with a low‑dimensional manifold improves accuracy, cuts function evaluations and speeds training convergence on benchmark datasets. Read more: https://getnews.me/manifold-constrained-neural-ode-improves-speed-and-accuracy/ #neuralode #manifoldlearning
Smooth activations like tanh and softplus give unique ODE solutions; ReLU keeps the NTK well‑conditioned, letting gradient descent achieve global convergence in Neural ODEs. Read more: https://getnews.me/neural-odes-gain-global-convergence-with-smooth-nonlinear-activations/ #neuralode #activations
September 29, 2025 at 3:11 PM
Smooth activations like tanh and softplus give unique ODE solutions; ReLU keeps the NTK well‑conditioned, letting gradient descent achieve global convergence in Neural ODEs. Read more: https://getnews.me/neural-odes-gain-global-convergence-with-smooth-nonlinear-activations/ #neuralode #activations
Zubov‑Net boosts classification accuracy while increasing robustness to noise and adversarial attacks in Neural ODE models; the research was submitted on 26 Sep 2025. Read more: https://getnews.me/zubov-net-boosts-neural-ode-accuracy-while-enhancing-robustness/ #neuralode #zubovnet #robustness
September 29, 2025 at 9:42 AM
Zubov‑Net boosts classification accuracy while increasing robustness to noise and adversarial attacks in Neural ODE models; the research was submitted on 26 Sep 2025. Read more: https://getnews.me/zubov-net-boosts-neural-ode-accuracy-while-enhancing-robustness/ #neuralode #zubovnet #robustness
With 90% of the training horizon, the physics-informed UDE achieved dramatically lower inventory forecast error than the fully data-driven NODE. Read more: https://getnews.me/ai-study-on-structured-vs-flexible-models-for-bullwhip-forecasting/ #bullwhipeffect #uede #neuralode
September 25, 2025 at 4:31 PM
With 90% of the training horizon, the physics-informed UDE achieved dramatically lower inventory forecast error than the fully data-driven NODE. Read more: https://getnews.me/ai-study-on-structured-vs-flexible-models-for-bullwhip-forecasting/ #bullwhipeffect #uede #neuralode
Researchers propose an iterative training algorithm for Neural ODEs using gradient descent in a Banach subspace, showing reduced performance loss under simulated disturbances. Read more: https://getnews.me/training-method-boosts-neural-odes-resistance-to-disturbances/ #neuralode #robustml
September 25, 2025 at 6:09 AM
Researchers propose an iterative training algorithm for Neural ODEs using gradient descent in a Banach subspace, showing reduced performance loss under simulated disturbances. Read more: https://getnews.me/training-method-boosts-neural-odes-resistance-to-disturbances/ #neuralode #robustml
RRIDs were included in this None paper. Thanks for making your methods matter! #ReproducibleResearch #methodsmatter #accelerateopenscience
CLADES: a hybrid NeuralODE-Gillespie approach for unveiling clonal cell fate and differentiation dynamics
doi.org
September 7, 2025 at 7:02 AM
RRIDs were included in this None paper. Thanks for making your methods matter! #ReproducibleResearch #methodsmatter #accelerateopenscience
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Recent lineage tracing single-cell techniques (LT-scSeq), e.g., the Lineage And RNA RecoverY (LARRY)
www.biorxiv.org
July 11, 2024 at 8:46 AM
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Super cool. Reminds me of the NeuralODE paper that motivates them as infinite-depth ResNets.
October 8, 2025 at 4:50 AM
Super cool. Reminds me of the NeuralODE paper that motivates them as infinite-depth ResNets.
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
August 27, 2025 at 10:15 AM
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Recent lineage tracing single-cell techniques (LT-scSeq), e.g., the Lineage And RNA RecoverY (LARRY)
www.biorxiv.org
July 11, 2024 at 8:46 AM
Unveiling Clonal Cell Fate and Differentiation Dynamics: A Hybrid NeuralODE-Gillespie Approach https://www.biorxiv.org/content/10.1101/2024.07.08.602444v1
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
August 27, 2025 at 10:15 AM
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
A new AI review! DiffEqML/torchdyn ⭐3.8/5.0
Torchdyn is a well-scoped PyTorch library for “numerical deep learning,” providing building blocks for neural differential equations (e.g., NeuralODE/NeuralSDE), implicit/deep equilibrium style models, and GPU-frie...
https://gitrated.com/DiffEqML/torchdyn
Torchdyn is a well-scoped PyTorch library for “numerical deep learning,” providing building blocks for neural differential equations (e.g., NeuralODE/NeuralSDE), implicit/deep equilibrium style models, and GPU-frie...
https://gitrated.com/DiffEqML/torchdyn
April 9, 2026 at 2:52 PM
A new AI review! DiffEqML/torchdyn ⭐3.8/5.0
Torchdyn is a well-scoped PyTorch library for “numerical deep learning,” providing building blocks for neural differential equations (e.g., NeuralODE/NeuralSDE), implicit/deep equilibrium style models, and GPU-frie...
https://gitrated.com/DiffEqML/torchdyn
Torchdyn is a well-scoped PyTorch library for “numerical deep learning,” providing building blocks for neural differential equations (e.g., NeuralODE/NeuralSDE), implicit/deep equilibrium style models, and GPU-frie...
https://gitrated.com/DiffEqML/torchdyn
VeloTrace Reconciles Divergent Velocity and Trajectory in Single-cell Transcriptomics with Deep Neural ODE [new]
A NeuralODE framework unifies divergent v & traj, learning a continuous velocity field tangent to cell fate trajs.
A NeuralODE framework unifies divergent v & traj, learning a continuous velocity field tangent to cell fate trajs.
April 13, 2026 at 5:14 PM
VeloTrace Reconciles Divergent Velocity and Trajectory in Single-cell Transcriptomics with Deep Neural ODE [new]
A NeuralODE framework unifies divergent v & traj, learning a continuous velocity field tangent to cell fate trajs.
A NeuralODE framework unifies divergent v & traj, learning a continuous velocity field tangent to cell fate trajs.
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models #NeuroDegeneration 🧪🧠
https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
August 27, 2025 at 4:02 PM
Predicting Alzheimer's Disease Progression from Sparse Multimodal Data by NeuralODE Models #NeuroDegeneration 🧪🧠
https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
https://www.biorxiv.org/content/10.1101/2025.08.26.672412v1
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
May 10, 2026 at 5:01 PM
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
March 3, 2026 at 2:41 PM
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
February 13, 2026 at 2:27 PM
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
August 26, 2025 at 4:06 AM
🌌 Ever heard of "Neural ODEs"? These clever algorithms blend traditional ODEs with neural networks, allowing models to learn continuous-time dynamics! It’s like teaching AI to dance through time! 🕺📈 #NeuralODE #AI #DeepLearning
ReQU networks, the stability and convergence rated for the RK scheme as well as metric entropy and concentration inequalities, we are able to prove that NeuralODE is a probably approximately correct (PAC) learning algorithm. [5/5 of https://arxiv.org/abs/2503.10729v1]
March 17, 2025 at 5:58 AM
ReQU networks, the stability and convergence rated for the RK scheme as well as metric entropy and concentration inequalities, we are able to prove that NeuralODE is a probably approximately correct (PAC) learning algorithm. [5/5 of https://arxiv.org/abs/2503.10729v1]
prior work, we do not only consider the statistical learning theory, but also give a detailed numerical analysis of the NeuralODE algorithm based on the 2nd order Runge-Kutta (RK) time integration. Using the universal approximation theory for deep [4/5 of https://arxiv.org/abs/2503.10729v1]
March 17, 2025 at 5:58 AM
prior work, we do not only consider the statistical learning theory, but also give a detailed numerical analysis of the NeuralODE algorithm based on the 2nd order Runge-Kutta (RK) time integration. Using the universal approximation theory for deep [4/5 of https://arxiv.org/abs/2503.10729v1]