Research map / Architectures and efficiency / RNNs and normalization
DevFly: Bio-Inspired Development of Binary Connections for Locality Preserving Sparse Codes
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace random PN-KC connections of FlyLSH with developmental wiring (Methods 1-3) | FlyLSH · MNIST | mAP: 9% for k = 32 to 106% for k = 2 | Helped |
| Developmental wiring on CIFAR-10 instead of random FlyLSH | DevFly · CIFAR-10 | mAP | Mixed |
| Add randomness to developmental wiring (Method 3) on GloVe | DevFly Method 3 · GloVe | mAP | No clear effect |
| Use DevFly instead of BioHash for training the projection | DevFly · MNIST | iterations to same precision: 10000 -> 320 | Helped |
| Feed samples ordered by label instead of random order (ablation) | DevFly Methods 1-3 · MNIST | mAP | No clear effect |
| Equalise sample norms before hashing (ablation) | DevFly · CIFAR-10 | mAP | Helped |
| Use sparse code with top-k WTA instead of dense code (DenseFly) for search | Method 1 · MNIST | search time | Helped |
| Enlarge feature space (larger m/k) | DevFly · four datasets | mAP advantage over FlyLSH | Hurt |
Closest papers by meaning
- Bio-Inspired Hashing for Unsupervised Similarity Search (ICML 2020 · similarity 0.90)
- Federated Nearest Neighbor Classification with a Colony of Fruit-Flies (AAAI 2022 · similarity 0.84)
- A Theoretical View on Sparsely Activated Networks (NeurIPS 2022 · similarity 0.84)
- Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery (NeurIPS 2024 · similarity 0.82)
- HOPSE: Scalable Higher-Order Positional and Structural Encoder for Combinatorial Representations (NeurIPS 2026 · similarity 0.82)
- What should a neuron aim for? Designing local objective functions based on information theory (ICLR 2025 · similarity 0.81)
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.