JoinBoost: Grow Trees Over Normalized Data Using Only SQL

Summary: JoinBoost compiles tree training into pure SQL over normalized joins, enabling factorized gradient boosting by treating Y as residuals on non-materialized joins using the variance semiring to support RMSE. Portable to any DBMS (DuckDB demos), it reduces residual-update costs via a residual projection column and outperforms prior systems—≈3× faster than LightGBM for RF and >10× vs prior In-DB ML—scaling to large schemas, many features, and complex join graphs. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID: 13147
Venue: VLDB
Year: 2023
Pagerank: 4.6120304e-05
Overall Rank: 7,920 | 44.96%
DOI: 10.14778/3611479.3611509

Incoming Non-self Citations Over Time

Authors

Incoming Citations (Sorted by Pagerank)

Showing 6 of 6 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
6,081	The Fast and the Private: Task-based Dataset Search	2024	CIDR	5.2179192e-05
6,377	Mitigating the Impedance Mismatch between Prediction Query Execution and Database Engine	2025	SIGMOD	5.0860948e-05
8,057	Biathlon: Harnessing Model Resilience for Accelerating ML Inference Pipelines	2024	VLDB	4.5903427e-05
9,855	In-Database Data Imputation	2024	SIGMOD	4.2652623e-05
10,177	InferF: Declarative Factorization of AI/ML Inferences over Joins	2026	SIGMOD	4.1905499e-05
10,580	Quantum Data Management in the NISQ Era	2025	VLDB	4.1905499e-05

Outgoing Citations (Sorted by Pagerank)

Showing 23 of 23 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
31	Provenance Semirings	2007	PODS	0.00078516827
139	The MADlib Analytics Library or MAD Skills, the SQL	2012	VLDB	0.00042320525
185	DuckDB: an Embeddable Analytical Database	2019	SIGMOD	0.00036529607
212	Join Synopses for Approximate Query Answering	1999	SIGMOD	0.00033997204
241	DB2 with BLU Acceleration: So Much More than Just a Column Store	2013	VLDB	0.00031314629
341	EmptyHeaded: A Relational Engine for Graph Processing	2016	SIGMOD	0.00026850764
419	Fast Serializable Multi-Version Concurrency Control for Main-Memory Database Systems	2015	SIGMOD	0.00023720294
539	Shark: SQL and Rich Analytics at Scale	2013	SIGMOD	0.00020615453
543	MLbase: A Distributed Machine-learning System	2013	CIDR	0.0002050918
557	SystemML: Declarative Machine Learning on Spark	2016	VLDB	0.00020186115
564	FAQ: Questions Asked Frequently	2016	PODS	0.00020002796
638	Towards a Unified Architecture for in-RDBMS Analytics	2012	SIGMOD	0.00018810785
832	Learning Linear Regression Models over Factorized Joins	2016	SIGMOD	0.00016089705
1,372	Random Sampling over Joins Revisited	2018	SIGMOD	0.0001233325
2,173	AJAR: Aggregations and Joins over Annotated Relations	2016	PODS	9.3767985e-05
3,009	On Functional Aggregate Queries with Additive Inequalities	2019	PODS	7.7230513e-05
3,280	A Layered Aggregate Engine for Analytics Workloads	2019	SIGMOD	7.2813732e-05
3,961	MLog: Towards Declarative In-Database Machine Learning	2017	VLDB	6.5824022e-05
4,200	Incremental View Maintenance with Triple Lock Factorization Benefits	2018	SIGMOD	6.3618329e-05
5,089	TCUDB: Accelerating Database with Tensor Processors	2022	SIGMOD	5.7017353e-05
5,952	PGMJoins: Random Join Sampling with Graphical Models	2021	SIGMOD	5.2547498e-05
9,694	Share the Tensor Tea: How Databases can Leverage the Machine Learning Ecosystem	2022	VLDB	4.2984337e-05
9,705	Distributed Numerical and Machine Learning Computations via Two-Phase Execution of Aggregated Join Trees	2021	VLDB	4.2951758e-05

Semantically Similar Papers

Overall Rank	Paper	Year	Venue	Pagerank
6,240	Scaling Similarity Joins over Tree-Structured Data	2015	VLDB	5.1362097e-05
4,717	Reducing Multidatabase Query Response Time By Tree Balancing	1995	SIGMOD	5.966866e-05
9,322	Are Joins over LSM-trees Ready? Take RocksDB as an Example	2025	VLDB	4.351469e-05
4,413	Robust Query Driven Cardinality Estimation under Changing Workloads	2023	VLDB	6.1989918e-05
7,180	Coresets over Multiple Tables for Feature-rich and Data-efficient Machine Learning	2023	VLDB	4.8032775e-05
901	To Join or Not to Join? Thinking Twice about Joins before Feature Selection	2016	SIGMOD	0.00015462938
832	Learning Linear Regression Models over Factorized Joins	2016	SIGMOD	0.00016089705
11,222	Lightweight Materialization for Fast Dashboards Over Joins	2023	SIGMOD	4.1905499e-05
9,471	DimBoost: Boosting Gradient Boosting Decision Tree to Higher Dimensions	2018	SIGMOD	4.3300949e-05
1,172	Learning Generalized Linear Models Over Normalized Data	2015	SIGMOD	0.00013504249