ML-LIRS: Leveraging Machine Learning to Improve the LIRS Replacement Algorithm

While the LIRS replacement algorithm is more capable at exploiting locality of block accesses than LRU by using its inter-reference recency (IRR) locality measure, it may still make mistakes in its decision-making thereby mis-evicting blocks from the cache. By leveraging machine learning techniques, LIRS can be improved to make more accurate decisions in situations where the IRR is untrustworthy. ...