An official website of the United States government
Abstract How do we ensure the veracity of science? The act of manipulating or fabricating scientific data has led to many high‐profile fraud cases and retractions. Detecting manipulated data, however, is a challenging and time‐consuming endeavor. Automated detection methods are limited due to the diversity of data types and manipulation techniques. Furthermore, patterns automatically flagged as suspicious can have reasonable explanations. Instead, we propose a nuanced approach where experts analyze tabular datasets, e.g., as part of the peer‐review process, using a guided, interactive visualization approach. In this paper, we present an analysis of how manipulated datasets are created and the artifacts these techniques generate. Based on these findings, we propose a suite of visualization methods to surface potential irregularities. We have implemented these methods in Ferret, a visualization tool for data forensics work. Ferret makes potential data issues salient and provides guidance on spotting signs of tampering and differentiating them from truthful data.
more »
« less
Anomaly Detection in Scientific Datasets using Sparse Representation
As the size and complexity of high-performance computing (HPC) systems keep growing, scientists' ability to trust the data produced is paramount due to potential data corruption for various reasons, which may stay undetected. While employing machine learning-based anomaly detection techniques could relieve scientists of such concern, it is practically infeasible due to the need for labels for volumes of scientific datasets and the unwanted extra overhead associated. In this paper, we exploit spatial sparsity profiles exhibited in scientific datasets and propose an approach to detect anomalies effectively. Our method first extracts block-level sparse representations of original datasets in the transformed domain. Then it learns from the extracted sparse representations and builds the boundary threshold between normal and abnormal without relying on labeled data. Experiments using real-world scientific datasets show that the proposed approach requires 13% on average (less than 10% in most cases and as low as 0.3%) of the entire dataset to achieve competitive detection accuracy (70.74%-100.0%) as compared to two state-of-the-art unsupervised techniques.
more »
« less
- PAR ID:
- 10441651
- Date Published:
- Journal Name:
- Proceedings of the First Workshop on AI for Systems
- Page Range / eLocation ID:
- 13 to 18
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
The emergence of data-intensive scientific discovery and machine learning has dramatically changed the way in which scientists and engineers approach materials design. Nevertheless, for designing macromolecules or polymers, one limitation is the lack of appropriate methods or standards for converting systems into chemically informed, machine-readable representations. This featurization process is critical to building predictive models that can guide polymer discovery. Although standard molecular featurization techniques have been deployed on homopolymers, such approaches capture neither the multiscale nature nor topological complexity of copolymers, and they have limited application to systems that cannot be characterized by a single repeat unit. Herein, we present, evaluate, and analyze a series of featurization strategies suitable for copolymer systems. These strategies are systematically examined in diverse prediction tasks sourced from four distinct datasets that enable understanding of how featurization can impact copolymer property prediction. Based on this comparative analysis, we suggest directly encoding polymer size in polymer representations when possible, adopting topological descriptors or convolutional neural networks when the precise polymer sequence is known, and using chemically informed unit representations when developing extrapolative models. These results provide guidance and future directions regarding polymer featurization for copolymer design by machine learning.more » « less
-
Neurosymbolic Programming (NP) techniques have the potential to accelerate scientific discovery. These models combine neural and symbolic components to learn complex patterns and representations from data, using high-level concepts or known constraints. NP techniques can interface with symbolic domain knowledge from scientists, such as prior knowledge and experimental context, to produce interpretable outputs. We identify opportunities and challenges between current NP models and scientific workflows, with real-world examples from behavior analysis in science: to enable the use of NP broadly for workflows across the natural and social sciences.more » « less
-
Scientific simulations generate large amounts of floating-point data, which are often not very compressible using the traditional reduction schemes, such as deduplication or lossless compression. The emergence of lossy floating-point compression holds promise to satisfy the data reduction demand from HPC applications; however, lossy compression has not been widely adopted in science production. We believe a fundamental reason is that there is a lack of understanding of the benefits, pitfalls, and performance of lossy compression on scientific data. In this paper, we conduct a comprehensive study on state-of- the-art lossy compression, including ZFP, SZ, and ISABELA, using real and representative HPC datasets. Our evaluation reveals the complex interplay between compressor design, data features and compression performance. The impact of reduced accuracy on data analytics is also examined through a case study of fusion blob detection, offering domain scientists with the insights of what to expect from fidelity loss. Furthermore, the trial and error approach to understanding compression performance involves substantial compute and storage overhead. To this end, we propose a sampling based estimation method that extrapolates the reduction ratio from data samples, to guide domain scientists to make more informed data reduction decisions.more » « less
-
Anomaly detection plays an important role in traffic operations and control. Missingness in spatial-temporal datasets prohibits anomaly detection algorithms from learning characteristic rules and patterns due to the lack of large amounts of data. This paper proposes an anomaly detection scheme for the 2021 Algorithms for Threat Detection (ATD) challenge based on Gaussian process models that generate features used in a logistic regression model which leads to high prediction accuracy for sparse traffic flow data with a large proportion of missingness. The dataset is provided by the National Science Foundation (NSF) in conjunction with the National Geospatial-Intelligence Agency (NGA), and it consists of thousands of labeled traffic flow records for 400 sensors from 2011 to 2020. Each sensor is purposely downsampled by NSF and NGA in order to simulate missing completely at random, and the missing rates are 99%, 98%, 95%, and 90%. Hence, it is challenging to detect anomalies from the sparse traffic flow data. The proposed scheme makes use of traffic patterns at different times of day and on different days of week to recover the complete data. The proposed anomaly detection scheme is computationally efficient by allowing parallel computation on different sensors. The proposed method is one of the two top performing algorithms in the 2021 ATD challenge.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1145/3588982.3603610
