NeuroCache is a modern, adaptive caching system that uses learning-based logic to optimize data retention and eviction. Unlike traditional caches like LRU (Least Recently Used) or LFU (Least Frequently Used) that follow static rules, NeuroCache dynamically learns and scores cached items based on their usage frequency and recency.
This intelligent behavior helps improve cache hit rates in scenarios like:
- Web browsers
- CPU/GPU memory caches
- CDN edge caching
- Recommendation systems
- Operating system page replacement
✅ Learns from usage patterns
✅ Eviction based on frequency and last-used time
✅ Lightweight and modular
✅ Clear output to visualize evictions and score changes
✅ Supports customization of scoring logic
NeuroCache assigns a score to each cache item:
score = frequency / (currentTime - lastUsedTime + 1)
- 📈 High frequency → more important
- ⏱️ Recently used → higher priority
- 🧮 Lower-scoring items get evicted when cache is full
This balances both recency and frequency, similar to a hybrid between LRU and LFU, with tunable logic for smarter decisions.
NeuroCache cache(3); // Set capacity to 3
cache.put(1, 100); // Add key 1
cache.put(2, 200); // Add key 2
cache.put(3, 300); // Add key 3
cache.get(1); // Access key 1 (boosts frequency)
cache.put(4, 400); // Cache full → evict based on lowest score
cache.get(2); // Key 2 might be evicted, will return -1 if soInitial insertions and access operations:
| Function | Time Complexity | Space Complexity | Description |
|---|---|---|---|
NeuroCache(int) |
O(1) | O(1) | Initializes cache capacity and timestamp. |
get(int key) |
O(1) average | O(1) | Retrieves value and updates metadata using a hash map. |
put(int key, int) |
O(1) average, O(n) worst | O(1) | Adds/updates key. Triggers eviction if cache is full. |
evict() |
O(n) | O(1) | Scans entries to evict one with the lowest score. |
computeScore() |
O(1) | O(1) | Calculates eviction score from frequency and recency. |
printCache() |
O(n) | O(1) | Displays all key-value pairs in the cache. |