arXiv subjects

Brian Hentschel

Publications and source records attributed to Brian Hentschel.

5 recordsLinked to original sources

arXiv2024

Exact PPS Sampling with Bounded Sample Size

Probability proportional to size (PPS) sampling schemes with a target sample size aim to produce a sample comprising a specified number $n$ of items while ensuring that each item in the population appears in the sample with a probability proportional to its specified "weight" (also called its "size"). These two objectives, however, cannot always be achieved simultaneously. Existing PPS schemes prioritize control of the sample size, violating the PPS property if necessary. We provide a new PPS scheme that allows a different trade-off: our method enforces the PPS property at all times while ensuring that the sample size never exceeds the target value $n$. The sample size is exactly equal to $n$ if possible, and otherwise has maximal expected value and minimal variance. Thus we bound the sample size, thereby avoiding storage overflows and helping to control the time required for analytics over the sample, while allowing the user complete control over the sample contents. The method is both simple to implement and efficient, being a one-pass streaming algorithm with an amortized processing time of $O(1)$ per item.

stat.ME↗

arXiv2020

MotherNets: Rapid Deep Ensemble Learning

Ensembles of deep neural networks significantly improve generalization accuracy. However, training neural network ensembles requires a large amount of computational resources and time. State-of-the-art approaches either train all networks from scratch leading to prohibitive training cost that allows only very small ensemble sizes in practice, or generate ensembles by training a monolithic architecture, which results in lower model diversity and decreased prediction accuracy. We propose MotherNets to enable higher accuracy and practical training cost for large and diverse neural network ensembles: A MotherNet captures the structural similarity across some or all members of a deep neural network ensemble which allows us to share data movement and computation costs across these networks. We first train a single or a small set of MotherNets and, subsequently, we generate the target ensemble networks by transferring the function from the trained MotherNet(s). Then, we continue to train these ensemble networks, which now converge drastically faster compared to training from scratch. MotherNets handle ensembles with diverse architectures by clustering ensemble networks of similar architecture and training a separate MotherNet for every cluster. MotherNets also use clustering to control the accuracy vs. training cost tradeoff. We show that compared to state-of-the-art approaches such as Snapshot Ensembles, Knowledge Distillation, and TreeNets, MotherNets provide a new Pareto frontier for the accuracy-training cost tradeoff. Crucially, training cost and accuracy improvements continue to scale as we increase the ensemble size (2 to 3 percent reduced absolute test error rate and up to 35 percent faster training compared to Snapshot Ensembles). We verify these benefits over numerous neural network architectures and large data sets.

cs.LG↗

arXiv2019

Temporally-Biased Sampling Schemes for Online Model Management

To maintain the accuracy of supervised learning models in the presence of evolving data streams, we provide temporally-biased sampling schemes that weight recent data most heavily, with inclusion probabilities for a given data item decaying over time according to a specified "decay function". We then periodically retrain the models on the current sample. This approach speeds up the training process relative to training on all of the data. Moreover, time-biasing lets the models adapt to recent changes in the data while---unlike in a sliding-window approach---still keeping some old data to ensure robustness in the face of temporary fluctuations and periodicities in the data values. In addition, the sampling-based approach allows existing analytic algorithms for static data to be applied to dynamic streaming data essentially without change. We provide and analyze both a simple sampling scheme (T-TBS) that probabilistically maintains a target sample size and a novel reservoir-based scheme (R-TBS) that is the first to provide both control over the decay rate and a guaranteed upper bound on the sample size. If the decay function is exponential, then control over the decay rate is complete, and R-TBS maximizes both expected sample size and sample-size stability. For general decay functions, the actual item inclusion probabilities can be made arbitrarily close to the nominal probabilities, and we provide a scheme that allows a trade-off between sample footprint and sample-size stability. The R-TBS and T-TBS schemes are of independent interest, extending the known set of unequal-probability sampling schemes. We discuss distributed implementation strategies; experiments in Spark illuminate the performance and scalability of the algorithms, and show that our approach can increase machine learning robustness in the face of evolving data.

cs.DB↗

arXiv2018

The Internals of the Data Calculator

Data structures are critical in any data-driven scenario, but they are notoriously hard to design due to a massive design space and the dependence of performance on workload and hardware which evolve continuously. We present a design engine, the Data Calculator, which enables interactive and semi-automated design of data structures. It brings two innovations. First, it offers a set of fine-grained design primitives that capture the first principles of data layout design: how data structure nodes lay data out, and how they are positioned relative to each other. This allows for a structured description of the universe of possible data structure designs that can be synthesized as combinations of those primitives. The second innovation is computation of performance using learned cost models. These models are trained on diverse hardware and data profiles and capture the cost properties of fundamental data access primitives (e.g., random access). With these models, we synthesize the performance cost of complex operations on arbitrary data structure designs without having to: 1) implement the data structure, 2) run the workload, or even 3) access the target hardware. We demonstrate that the Data Calculator can assist data structure designers and researchers by accurately answering rich what-if design questions on the order of a few seconds or minutes, i.e., computing how the performance (response time) of a given data structure design is impacted by variations in the: 1) design, 2) hardware, 3) data, and 4) query workloads. This makes it effortless to test numerous designs and ideas before embarking on lengthy implementation, deployment, and hardware acquisition steps. We also demonstrate that the Data Calculator can synthesize entirely new designs, auto-complete partial designs, and detect suboptimal design choices.

cs.DB↗

arXiv2018

Temporally-Biased Sampling for Online Model Management

To maintain the accuracy of supervised learning models in the presence of evolving data streams, we provide temporally-biased sampling schemes that weight recent data most heavily, with inclusion probabilities for a given data item decaying exponentially over time. We then periodically retrain the models on the current sample. This approach speeds up the training process relative to training on all of the data. Moreover, time-biasing lets the models adapt to recent changes in the data while -- unlike in a sliding-window approach -- still keeping some old data to ensure robustness in the face of temporary fluctuations and periodicities in the data values. In addition, the sampling-based approach allows existing analytic algorithms for static data to be applied to dynamic streaming data essentially without change. We provide and analyze both a simple sampling scheme (T-TBS) that probabilistically maintains a target sample size and a novel reservoir-based scheme (R-TBS) that is the first to provide both complete control over the decay rate and a guaranteed upper bound on the sample size, while maximizing both expected sample size and sample-size stability. The latter scheme rests on the notion of a "fractional sample" and, unlike T-TBS, allows for data arrival rates that are unknown and time varying. R-TBS and T-TBS are of independent interest, extending the known set of unequal-probability sampling schemes. We discuss distributed implementation strategies; experiments in Spark illuminate the performance and scalability of the algorithms, and show that our approach can increase machine learning robustness in the face of evolving data.

cs.DB↗