Machine learning to extract information from noise - application to concrete and cancer detection

Aug 3, 2026

Uploaded by Noah

Availability of large experimental and computational datasets, together with powerful computing, has enabled machine learning to forecast physical-science quantities. Yet real measurements contain noise that propagates into model training and limits prediction accuracy and uncertainty, restricting broader applicability. This thesis develops a machine learning methodology that extracts information from noise, converting uncertainty from a liability into a productive signal for improved inference. Results are demonstrated for concrete mix design and long-term strength, and for early acute myeloid leukemia detection from DNA methylation patterns.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages152

Reading Time6 h 23 min

AccessFree to download

Abstract

Core challenge: noise in real-world data

Methodology: extracting information from noise

Applications in concrete

Applications in cancer detection

Broader applicability and impact

Machine learning to extract information from noise: application to concrete and cancer detection Bahdan Zviazhynski Supervisor: Dr Gareth Conduit Department of Physics University of Cambridge This thesis is submitted for the degree of Doctor of Philosophy Trinity College July 2024 This thesis is the result of my own work and includes nothing which is the outcome of work done in collaboration except as declared in the preface and speciﬁed in the text. It is not substantially the same as any work that has already been submitted, or is being concurrently submitted, for any degree, diploma or other qualiﬁcation at the University of Cambridge or any other University or similar institution except as declared in the preface and speciﬁed in the text. It does not exceed the prescribed word limit for the relevant Degree Committee. To my parents. Abstract The availability of large amounts of experimental and computational data together with powerful computers enabled use of machine learning to predict phenomena in physical sciences. Machine learning is a class of methods that start from existing data to train a model and then use this model to predict the quantities of interest. While machine learning has been successfully applied in physical sciences and beyond, it faces a challenge presented by noise in real-world experimental data. This noise leads to inaccuracy and uncertainty in machine learning predictions, limiting applications of machine learning to physical sciences. In this thesis, we develop a machine learning methodology that extracts information from noise in data, turning the negative impact of noise and uncertainty into an asset for making better predictions. We demonstrate the effectiveness of the methodology in two research areas: concrete and cancer detection. First, we utilize the methodology in two studies of concrete. In one study, we propose two concrete mixes that were experimentally veriﬁed to exhibit improved environmental cost and carbonation. In the other study, we make predictions of long-term concrete strength that were experimentally veriﬁed. Second, we utilize the methodology to extract information from the noise in DNA methylation patterns to detect acute myeloid leukemia – a type of blood cancer – at early stages. We achieve area under receiver operating characteristic curve of above 0.8 for new, previously unseen patients. This means we can correctly distinguish between a healthy and an unhealthy patient prior to the onset of the disease in more than 80% of the cases. Our methodology is generic and can be applied in many ﬁelds, including information engineering, autonomous vehicles, and additive manufacturing, where the information embedded in noise can be exploited to accelerate development, understanding, and impact. Preface This thesis describes the work done between October 2021 and July 2024 in the Theory of Condensed Matter (TCM) group at the Cavendish Laboratory, Cambridge, under the supervision of Dr Gareth Conduit. Chapter 1 introduces the core concepts and ideas used in this thesis. Subsequent chapters contain original material that is in preparation for submission or is published/under review elsewhere as follows: Chapter 2: Zviazhynski, B. & Conduit, G. Unveil the unseen: Exploit information hidden in noise. Applied Intelligence 53, 11966–11978 (2023) . [251] Chapter 3: Zviazhynski, B. & Conduit, G. Combined metric: coefﬁcient of determination meets quality of uncertainty. Under review with Stat journal. Chapter 4: Forsdyke, J., Zviazhynski, B., Lees, J., & Conduit, G. Probabilistic selection and design of concrete using machine learning. Data Centric Engineering 4, 1–18 (2023) . [68] This thesis is the result of my own work and includes nothing which is the outcome of work done in collaboration except as declared in the Preface, Acknowledgements, and as speciﬁed in the text. It is not substantially the same as any work that has already been submitted

FAQs

What problem does the thesis address in machine learning?

Noise in real experimental data creates inaccuracy and uncertainty that limits machine-learning predictions in physical sciences.

How does the thesis improve predictions when data are noisy?

It develops a methodology that extracts information from noise, turning the negative impact of uncertainty into an asset for better inference.

Where is the methodology validated?

The thesis demonstrates effectiveness in two domains: concrete mix design and long-term concrete strength prediction, and early acute myeloid leukemia detection using DNA methylation patterns.

Aug 3, 2026

Uploaded by Noah