Research map / Probabilistic and scientific modeling / Dynamical systems
Factorized Neural Processes for Neural Processes: K-Shot Prediction of Neural Responses
Ronald (James) Cotton, Fabian H. Sinz, Andreas Tolias
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 a factorized latent space for receptive-field location and tuning properties | Neural Processes for neural response prediction · Simulated neural responses | predictive accuracy | Improved |
| Increase the number of observed trials | Factorized Neural Process · Simulated and mouse V1 neural responses | predictive accuracy | Improved |
| Train with maximum set size 1024 rather than 512 trials (ablation) | Factorized Neural Process · Simulated neural responses | asymptotic predictive performance | Improved |
| Increase tuning-function property latent dimension (ablation) | Factorized Neural Process · Simulated neural responses | predictive performance | Improved |
| Randomize phase offset in simulated receptive fields (ablation) | Factorized Neural Process · Simulated neural responses | predictive accuracy | Hurt |
| Include complex cells along with simple cells (ablation) | Factorized Neural Process · Simulated neural responses | predictive performance | No clear effect |
| Replace the softmax spatial-statistic Gaussian approximation with an MLP mapping to mean and standard deviation (ablation) | Factorized Neural Process location encoder · Simulated neural responses | training speed | Hurt |
| Use a Group CNN DenseNet rather than a standard CNN (ablation) | Factorized Neural Process feature extractor · Visual response prediction | training reliability and speed | Improved |
| Use an MLP transformation of the tuning-properties latent in the decoder (ablation) | Factorized Neural Process conditional decoder · Visual response prediction | predictive performance | Improved |
| Condition predictions on a learned prior over tuning functions instead of optimizing each neuron’s readout | Factorized Neural Process versus PNO · Mouse V1, fewer than 500 trials | predictive accuracy | Improved |
| Infer the tuning function with a single forward pass instead of per-neuron optimization | Factorized Neural Process versus PNO · Mouse V1 | inference time: 250~ms vs $\sim20$~s | Improved |
Closest papers by meaning
- Modeling Dynamic Neural Activity by combining Naturalistic Video Stimuli and Stimulus-independent Latent Factors (NeurIPS 2025 · similarity 0.88)
- Reproducibility of predictive networks for mouse visual cortex (NeurIPS 2024 · similarity 0.85)
- Neuroformer: Multimodal and Multitask Generative Pretraining for Brain Data (ICLR 2024 · similarity 0.85)
- Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex. (NeurIPS 2024 · similarity 0.85)
- Self-Attention-Based Contextual Modulation Improves Neural System Identification (ICLR 2025 · similarity 0.84)
- Retrospective for the Dynamic Sensorium Competition for predicting large-scale mouse primary visual cortex activity from videos (NeurIPS 2024 · similarity 0.84)
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.
