Predictive Caching in Non-Stationary Environments: A Time Series Prediction and Survival Analysis Approach
Bibliographic record
Abstract
This paper introduces an innovative predictive caching strategy tailored to a real-world dataset, specifically the Facebook video dataset. Making caching decisions for the dataset is challenging due to its dynamic nature, where users’ content requests vary over time without fitting into any known models. Traditional caching strategies, which often rely on a constant pool of files, do not suit this dataset as content is requested by users, and then its popularity fades over time; furthermore, the list of available content changes. We propose a two-stage predictive caching strategy. Initially, it forecasts the number of user requests using content features and historical request data, achieved through training a long short-term memory (LSTM) network. Then, we employ our proposed extended Cox proportional hazard (E-CPH) model to predict the survival probability of content. This facilitates proactive content caching. Caching new content is made possible by the timely eviction of content unlikely to be requested again. To incorporate the predicted content popularity and its life cycle into the caching decision, we introduce a partially observable Markov decision process (POMDP)-based caching strategy. Here, the survival probability of content contributes to the belief state of the associated content which leads to our believed predicted reward - a cache hit. The caching algorithm then stores the files based on their predicted believed reward taking into account both the popularity and survival probability predictions. Simulation results validate the efficacy of our proposed predictive caching method in enhancing the cache hit rate compared to conventional recurrent neural network (RNN)-based caching and policy-based caching approaches, such as least frequently used caching and its variants.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".