Biathlon: Harnessing Model Resilience for Accelerating ML Inference Pipelines

Summary: Biathlon, an ML-serving system, exploits model resilience to input perturbations by selecting per-aggregation-feature approximation levels to maximize latency reduction while guaranteeing bounded end-to-end accuracy loss. Evaluated on real pipelines, it achieves 5.3x–16.6x speedups with negligible accuracy drop. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID: h545ab1ee6c8e6be8
Venue: VLDB
Year: 2024
Pagerank: 5.402957e-05
Overall Rank: 7,950 | 46.72%
DOI: 10.14778/3675034.3675052
PDF: Download (CC BY-NC-ND 4.0)

Incoming Non-self Citations Over Time

Authors

1. Chaokun Chang (Chinese University of Hong Kong)
2. Eric Lo (Chinese University of Hong Kong)
3. Chunxiao Ye (Chinese University of Hong Kong)

BibTeX Citation

@article{chang_vldb24,
        title = {{Biathlon: Harnessing Model Resilience for Accelerating ML Inference Pipelines}},
        author = {Chang, Chaokun and Lo, Eric and Ye, Chunxiao},
        journal = {PVLDB},
        series = {{VLDB} '24},
        volume = {17},
        number = {10},
        pages = {2631--2640},
        doi = {10.14778/3675034.3675052},
        url = {https://doi.org/10.14778/3675034.3675052},
        year = {2024}
}

Incoming Citations (Sorted by Pagerank)

Showing 3 of 3 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
4,910	Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving	2025	SIGMOD	6.3468554e-05
11,139	KEN: An Execution Engine for Unstructured Database Systems	2026	VLDB	4.9605343e-05
11,257	Malleus: Straggler-Resilient Hybrid Parallel Training of Large-scale Models via Malleable Data and Model Parallelization	2025	SIGMOD	4.9605343e-05

Outgoing Citations (Sorted by Pagerank)

Showing 27 of 27 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
9	Online Aggregation	1997	SIGMOD	0.00076079812
273	NoScope: Optimizing Neural Network Queries over Video at Scale	2017	VLDB	0.00022518449
320	DeepDB: Learn from Data, not from Queries!	2020	VLDB	0.00021128745
594	Wander Join: Online Aggregation via Random Walks	2016	SIGMOD	0.00015814493
770	VerdictDB: Universalizing Approximate Query Processing	2018	SIGMOD	0.00014086699
841	Quickr: Lazily Approximating Complex AdHoc Queries in BigData Clusters	2016	SIGMOD	0.00013503124
1,435	Knowing When You’re Wrong: Building Fast and Reliable Approximate Query Processing Systems	2014	SIGMOD	0.00010659454
1,839	DBEst: Revisiting Approximate Query Processing Engines with Machine Learning Models	2019	SIGMOD	9.5187688e-05
1,874	G-OLA: Generalized On-Line Aggregation for Interactive Analysis on Big Data	2015	SIGMOD	9.4470874e-05
2,014	Sample + Seek: Approximating Aggregates with Distribution Precision Guarantee	2016	SIGMOD	9.1775869e-05
2,040	Database Learning: Toward a Database that Becomes Smarter Every Time	2017	SIGMOD	9.1319608e-05
2,199	Extending Relational Query Processing with ML Inference	2020	CIDR	8.8664379e-05
2,258	Evaluating End-to-End Optimization for Data Analytics Applications in Weld	2018	VLDB	8.7387151e-05
2,464	Query Processing on Tensor Computation Runtimes	2022	VLDB	8.4081954e-05
2,672	End-to-end Optimization of Machine Learning Prediction Queries	2022	SIGMOD	8.1301225e-05
2,682	Relational Confidence Bounds Are Easy With The Bootstrap*	2005	SIGMOD	8.1203255e-05
3,220	Turbo-Charging Estimate Convergence in DBO	2009	VLDB	7.5077984e-05
3,415	A Demonstration of Willump: A Statistically-Aware End-to-end Optimizer for Machine Learning Inference	2020	VLDB	7.3054307e-05
3,513	Serving and Optimizing Machine Learning Workflows on Heterogeneous Infrastructures	2023	VLDB	7.2253231e-05
4,198	Rafiki: Machine Learning as an Analytics Service System	2019	VLDB	6.7303243e-05
6,049	Incremental Computation of Common Windowed Holistic Aggregates	2016	VLDB	5.8866676e-05
6,249	Optimizing In-memory Database Engine for AI-powered On-line Decision Augmentation Using Persistent Memory	2021	VLDB	5.8190376e-05
6,428	Containerized Execution of UDFs: An Experimental Evaluation	2022	VLDB	5.7709339e-05
6,562	JoinBoost: Grow Trees Over Normalized Data Using Only SQL	2023	VLDB	5.7287508e-05
9,213	FEBench: A Benchmark for Real-Time Relational Data Feature Extraction	2023	VLDB	5.1926696e-05
10,150	Everest: A Top-K Deep Video Analytics System	2022	SIGMOD	5.0597449e-05
10,164	RALF: Accuracy-Aware Scheduling for Feature Store Maintenance	2024	VLDB	5.0570156e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	10,040	cedar: Optimized and Unified Machine Learning Input Data Pipelines	2025	VLDB
2	4,910	Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving	2025	SIGMOD
3	948	Democratizing Data Science through Interactive Curation of ML Pipelines	2019	SIGMOD
4	9,196	Optimizing Inference Serving on Serverless Platforms	2022	VLDB
5	9,454	Towards Observability for Production Machine Learning Pipelines	2022	VLDB
6	6,765	Mitigating the Impedance Mismatch between Prediction Query Execution and Database Engine	2025	SIGMOD
7	5,793	Optimizing Data Pipelines for Machine Learning in Feature Stores	2023	VLDB
8	6,234	Materialization and Reuse Optimizations for Production Data Science Pipelines	2022	SIGMOD
9	5,291	InferDB: In-Database Machine Learning Inference Using Indexes	2024	VLDB
10	7,558	Automating and Optimizing Data-Centric What-If Analyses on Native Machine Learning Pipelines	2023	SIGMOD