Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Deep Random Splines for Point Process Intensity Estimation of Neural Population Data
Gabriel Loaiza-Ganem, Sean Perkins, Karen Schroeder, Mark Churchland, John P. Cunningham
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 |
|---|---|---|---|
| Use DRS-VAE instead of PfLDS | Point-process latent representation and intensity estimation · Simulated data | ELBO: 52.3 -> 57.1 | Improved |
| Use DRS-VAE instead of PP-GPFA | Point-process latent representation and intensity estimation · Simulated data | L² distance: 15.4 ± 9.64 -> 4.43 ± 3.55 | Improved |
| Use DRS-VAE instead of PfLDS | Point-process latent representation and intensity estimation · Reaching data | 15-NN accuracy: 3.1% -> 23.7% | Improved |
| Use DRS-VAE instead of PfLDS | Point-process latent representation and intensity estimation · Reaching data | ELBO: -505.7 -> -500.8 | Improved |
| Use DRS-VAE instead of PfLDS | Point-process latent representation and intensity estimation · Cycling data | ELBO: 6532 -> 6372 | Hurt |
| Use DRS-VAE instead of PfLDS | Point-process latent representation and intensity estimation · Cycling data | 15-NN accuracy: 11.7% -> 55.9% | Improved |
| Use DRS-VAE instead of PP-GPFA | Point-process latent representation and intensity estimation · Cycling data | SSG/SST: 14.6% -> 70.0% | Improved |
| Use squared splines as nonnegative intensities instead of the constrained spline parameterization (ablation) | DRS-VAE · Point-process intensity modeling | Performance | Hurt |
| Use nonnegative-coefficient B-spline basis instead of alternating projections (ablation) | DRS-VAE · Point-process intensity modeling | Performance | Hurt |
| Use bidirectional LSTMs instead of regular LSTMs in the encoder (ablation) | DRS-VAE · Point-process encoder | Performance | No clear effect |
| Increase the number of spline intervals (ablation) | DRS-VAE · Cycling data | ELBO | No clear effect |
| Increase the number of time bins (ablation) | PfLDS · Cycling data | ELBO: increased slightly | Improved |
Closest papers by meaning
- Enabling hyperparameter optimization in sequential autoencoders for spiking neural data (NeurIPS 2019 · similarity 0.86)
- A Differentiable Point Process with Its Application to Spiking Neural Networks (ICML 2021 · similarity 0.86)
- Exact, Fast and Expressive Poisson Point Processes via Squared Neural Families (AAAI 2024 · similarity 0.85)
- Latent Diffusion for Neural Spiking Data (NeurIPS 2024 · similarity 0.84)
- P-SPIKESSM: HARNESSING PROBABILISTIC SPIKING STATE SPACE MODELS FOR LONG-RANGE DEPENDENCY TASKS (ICLR 2025 · similarity 0.84)
- Noise-Injected Spiking Graph Convolution for Energy-Efficient 3D Point Cloud Denoising (AAAI 2025 · similarity 0.83)
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.
