Research map / Learning theory and algorithms / Matrix and tensor methods
ReAl-LiFE: Accelerating the Discovery of Individualized Brain Connectomes on GPUs
Sawan Kumar, Varsha Sreenivasan, Partha Talukdar, Franco Pestilli, Devarajan Sridharan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| run LiFE on GPU instead of CPU | LiFE · dMRI dataset S, 2 million fibers | speedup: 98x | Improved |
| increase connectome size from small to 3 million fibers | ReAl-LiFE GPU · dMRI dataset I | speedup: ~80x | Improved |
| increase connectome size to 1 million fibers | ReAl-LiFE GPU · dMRI dataset H | speedup: 63x | Improved |
| use GPU acceleration instead of MPI parallelization | LiFE · Cluster A configuration | speedup: 8.7x -> 98x | Improved |
| add L1 regularization (lambda 0.01) to LiFE pruning (ablation) | LiFE · dMRI dataset I | cross-validation RMSE: 0.936 -> 0.922 | Improved |
| add L1 regularization (lambda 0.01) to LiFE pruning (ablation) | LiFE · dMRI dataset S | cross-validation RMSE: 0.854 -> 0.843 | Improved |
| use larger lambda up to 0.2 (up to 80% fewer connections) (ablation) | ReAl-LiFE · dMRI datasets I and S | cross-validation RMSE | Improved |
| prune ensemble connectome with ReAl-LiFE instead of LiFE (ablation) | LiFE · dMRI dataset S ensemble tractography | cross-validation RMSE: 22% fewer connections | Improved |
| prune ensemble connectome with ReAl-LiFE instead of LiFE (ablation) | LiFE · dMRI dataset S ensemble tractography | non-zero weight fibers (million): 0.207 -> 0.232 | Mixed |
| use ReAl-LiFE weights instead of LiFE weights as SVM features (ablation) | SVM classifier · ADNI AD vs NC | accuracy: 65.5% -> 78.8% | Improved |
| use ReAl-LiFE weights instead of fiber counts (ablation) | SVM classifier · ADNI AD vs NC | accuracy: 66.1% -> 78.8% | Improved |
| use ReAl-LiFE weights instead of SIFT2 weights (ablation) | SVM classifier · ADNI AD vs NC | accuracy: 65.9% -> 78.8% | Improved |
Closest papers by meaning
- Unified representation of tractography and diffusion-weighted MRI data using sparse multidimensional arrays (NeurIPS 2017 · similarity 0.88)
- MAGNET: Manifold-Aware Graph Diffusion Network for Connectome Generation (NeurIPS 2026 · similarity 0.85)
- Optimal Brain Apoptosis (ICLR 2025 · similarity 0.82)
- $\textit{NeuroPath}$: A Neural Pathway Transformer for Joining the Dots of Human Connectomes (NeurIPS 2024 · similarity 0.82)
- Neurodegenerative Brain Network Classification via Adaptive Diffusion with Temporal Regularization (ICML 2024 · similarity 0.82)
- ConnectomeBench: Can LLMs proofread the connectome? (NeurIPS 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
