gdymind / Awesome-learned-index

A set of learned index papers w/o notes

Awesome-learned-index

This is a collection of learned index papers with notes.

By year

2018

1. 2018-The case for learned index: the first learned index paper. It introduces RMI that use tree-like structures. Inner nodes are models while leaf nodes are data. Each model maps from a key $k$ to a position $p$($k\rightarrow p$).

2019

1. 2019-SOSD_Benchmark

2. 2019-aiDM-Considerations_for_Updates

3. 2019-arxiv-Scalable_Learned_Index_in_Storage: the initial version of FINEdex

4. 2019-SIGMOD-FITing-Tree_Data-aware-Index

2020

1. 2020-aiDM-Radix_Spline: Using linear spine fits to a CDF, then a flat radix table as an appoximate index.
2. 2020-APSys-SIndex_Scalable_Learned_Index__String_Keys
3. 2020-ICDEW-SMART-Self_Tuning_ART
4. 2020-OSDI-Bourbon_learned_LSM
5. 2020-OSDI-Bourbon_learned_LSM_slides
6. 2020-PPoPP-XIndex_Scalable_Learned_Index_for_Multicore_Data_Storage
7. 2020-SIGMOD-ALEX_Updatable_Adaptive_Learned_Index
8. 2020-SIGMOD-CDFShop-Exploring_and_Optimizing_Learned_Index_Structures: tuning parameters of RMIs
9. 2020-VLDB-PGM-index_fully-dynamic_compressed_worst-case_bounds
10. 2020-workshop_NIPS_Learned_Index_for_bigtable
11. 2020-SIGMOD-HOPE: not learned index, but an encoding schme; order persevering encoding for string; can be used for string learned indexes
12. 2020-SIGMOD_The_Case_for_a_Learned_Sorting_Algorithm

2021

1. 2021-PVLDB-Benchmarking_Learned_Indexes
2. 2021-AIDB-PLEX_RS+CHT
3. 2021-AIDB-RSS_Bounding_the_Last_Mile-Efficient_Learned_String_Indexing
4. 2021-PVLDB-FINEdex-Fine-grained_for_Scalable_Concurrent_Memory_Systems
5. 2021-PVLDB-LIPP_Updatable_Learned_Index_Precise_Positions
6. 2021-aiDM-RUSLI_Real-time_Updatable_Spline_Learned_Index
7. 2021-aiDM-Tailored_Regression_Learned_Indexes-Logarithmic-Error-Regression
8. 2021-PVLDB-Endownment-learned_cardition

2022

1. 2022-PVLDB-are_updatable_learned_index_ready
2. 2022-ICLR_learned_index_with_dynamic_eps
3. 2022-VLDB-NFL_Learned_Index_Distribution_Transformation: it transforms keys distribution to make it more linear. source code
4. 2022-TOS-Xindex-most-recent
5. 2022-VLDB-APEX_Learned_Index_PM
6. 2022-learned_Similarity_Search
7. 2022-PVLDB-Endowment-FILM-for-Larger-than-Memory-Databases
8. 2022-PVLDB-Endowment-Can_Learned_Models_Replace_Hash_Functions

2023

1. 2023-FAST-ROLEX
2. 2023-ASPLOS-LeaFTL- A Learning-Based Flash Translation Layer for Solid-State Drives Learned index for SSD FTL page-level memory mapping
3. 2023-PVLDB-Endowment-Comprehensive_Experimental_Evaluation
4. 2023-arxiv-Updatable_Learned_Indexes_Disk-Resident_DBMS
5. 2023-arxiv-DILI-A Distribution-Driven Learned Index Using more bulk loading time for better lookup.

By category

Survey/Benchmark/Tuning

1. 2019-SOSD_Benchmark
2. 2020-SIGMOD-demo-CDFShop-tuning_RMI
3. 2021-PVLDB-Benchmarking_Learned_Indexes
4. 2021-aiDM-Tailored_Regression_Learned_Indexes-Logarithmic-Error-Regression
5. 2022-are_updatable_learned_index_ready
6. 2023-PVLDB-Endowment-Comprehensive_Experimental_Evaluation

Read-only

1. The case for learned index
2. 2020-aiDM-Radix_Spline
3. 2020-workshop_NIPS_Learned_Index_for_bigtable
4. 2021-AIDB-PLEX_RS+CHT: RadixSpine as the top + Compact Hist-Tree as the bottom
5. 2021-AIDB-RSS_Bounding_the_Last_Mile-Efficient_Learned_String_Indexing

Updatable

1. 2019-SIGMOD-FITing-Tree_Data-aware-Index
2. 2020-SIGMOD-ALEX_Updatable_Adaptive_Learned_Index Use gapped array for SMO
3. 2020-VLDB-PGM-index_fully-dynamic_compressed_worst-case_bounds
4. 2021-PVLDB-LIPP_Updatable_Learned_Index_Precise_Positions
5. 2021-aiDM-RUSLI_Real-time_Updatable_Spline_Learned_Index
6. 2022-TOS-Xindex-most-recent
7. 2023-FAST-ROLEX
8. 2023-arxiv-DILI-A Distribution-Driven Learned Index Using more bulk loading time for better lookup.

Secondary Storage/Persistent Memory/LSM

1. 2019-arxiv-Scalable_Learned_Index_in_Storage: the initial version of FINEdex
2. 2020-workshop_NIPS_Learned_Index_for_bigtable
3. 2020-OSDI-Bourbon_learned_LSM
4. 2020-OSDI-Bourbon_learned_LSM_slides
5. 2022-aiDM-LSI-Learned_Secondary_Index_Structure
6. 2022-VLDB-APEX_Learned_Index_PM
7. 2022-PVLDB-Endowment-FILM-for-Larger-than-Memory-Databases
8. 2023-arxiv-Updatable_Learned_Indexes_Disk-Resident_DBMS

Radix-Spine based

1. 2020-aiDM-Radix_Spline: Using linear spine fits to a CDF, then a flat radix table as an appoximate index.
2. 2021-AIDB-RSS_Bounding_the_Last_Mile-Efficient_Learned_String_Indexing
3. 2021-AIDB-PLEX_RS+CHT
4. 2021-aiDM-RUSLI_Real-time_Updatable_Spline_Learned_Index

Variable length string keys

1. 2020-APSys-SIndex_Scalable_Learned_Index__String_Keys
2. 2021-AIDB-RSS_Bounding_the_Last_Mile-Efficient_Learned_String_Indexing
3. 2020-SIGMOD-HOPE: not learned index, but an encoding schme; order persevering encoding for string; can be used for string learned indexes
4. 2020-SIGMOD-HOPE_slides

Concurrency

1. 2020-PPoPP-XIndex_Scalable_Learned_Index_for_Multicore_Data_Storage
2. 2020-APSys-SIndex_Scalable_Learned_Index_String_Keys
3. 2021-PVLDB-FINEdex-Fine-grained_for_Scalable_Concurrent_Memory_Systems
4. 2022-TOS-Xindex-most-recent

Applications

1. 2020-SIGMOD_The_Case_for_a_Learned_Sorting_Algorithm
2. 2022-learned_Similarity Search
3. 2022-PVLDB-Endowment-Can_Learned_Models_Replace_Hash_Functions
4. 2022-learned_Similarity_Search
5. 2023-ASPLOS-LeaFTL-Learning-Based Flash Translation Layer for Solid-State Drives Learned index for SSD FTL page-level memory mapping