A5091107483- Machine Learning and Data Classification
- Neural Networks and Applications
- Evolutionary Algorithms and Applications
- Face and Expression Recognition
- Anomaly Detection Techniques and Applications
- Adversarial Robustness in Machine Learning
- Statistical Methods and Inference
- Parallel Computing and Optimization Techniques
- Machine Learning and Algorithms
- Metaheuristic Optimization Algorithms Research
- Data Stream Mining Techniques
- Advanced Statistical Methods and Models
- Gene expression and cancer classification
- Imbalanced Data Classification Techniques
- Bayesian Modeling and Causal Inference
- Data Mining Algorithms and Applications
- Viral Infectious Diseases and Gene Expression in Insects
- Domain Adaptation and Few-Shot Learning
- Time Series Analysis and Forecasting
- Rough Sets and Fuzzy Logic
- Natural Language Processing Techniques
- Advanced Memory and Neural Computing
- Distributed systems and fault tolerance
- Software System Performance and Reliability
- Advanced Neural Network Applications
University of Manchester
2015-2024
Universidad de Sevilla
2022
Boston University
2017-2021
Fulbourn Hospital
2020
Science Oxford
2020
Technische Universität Berlin
2012
Intel (United States)
2010
Defence Science and Technology Laboratory
2009
Manchester University
2007
University of Birmingham
2003-2004
Ensembles are a widely used and effective technique in machine learning---their success is commonly attributed to the degree of disagreement, or 'diversity', within ensemble. For ensembles where individual estimators output crisp class labels, this 'diversity' not well understood remains an open research issue. regression estimators, diversity can be exactly formulated terms covariance between estimator outputs, optimum level expressed bias-variance-covariance trade-off. Despite this, most...
Learning in adversarial settings is becoming an important task for application domains where attackers may inject malicious data into the training set to subvert normal operation of data-driven technologies. Feature selection has been widely used machine learning security applications improve generalization and computational efficiency, although it not clear whether its use be beneficial or even counterproductive when are poisoned by intelligent attackers. In this work, we shed light on...
The identification of biomarkers to support decision-making is central personalized medicine, in both clinical and research scenarios. challenge can be seen two halves: identifying predictive markers, which guide the development/use tailored therapies; prognostic other aspects care trial planning, i.e. markers considered as covariates for stratification. Mistakenly assuming a biomarker predictive, when it fact largely (and vice-versa) highly undesirable, result financial, ethical personal...
Deep neural networks have been widely adopted in recent years, exhibiting impressive performances several application domains. It has however shown that they can be fooled by adversarial examples, i.e., images altered a barely-perceivable noise, carefully crafted to mislead classification. In this work, we aim evaluate the extent which robot-vision systems embodying deep-learning algorithms are vulnerable and propose computationally efficient countermeasure mitigate threat, based on...
Abstract The Domestic Abuse, Stalking and Honour Based Violence (DASH) form is a standardized risk assessment implemented across most UK police forces. It intended to facilitate an officer’s structured professional judgment about the victim faces of serious harm at hand their abuser. Until now, it has been open question whether this tool works in practice. Here, we present largest scale European study, making case that underperforming. Each element DASH questionnaire is, best, weakly...
We provide a unifying perspective for two decades of work on cost-sensitive Boosting algorithms. When analyzing the literature 1997–2016, we find 15 distinct variants original algorithm; each these has its own motivation and claims to superiority—so who should believe? In this critique using four theoretical frameworks: Bayesian decision theory, functional gradient descent view, margin probabilistic modelling. Our finding is that only three algorithms are fully supported—and model view...
We present a theory of ensemble diversity, explaining the nature diversity for wide range supervised learning scenarios. This challenge has been referred to as holy grail learning, an open research issue over 30 years. Our framework reveals that is in fact hidden dimension bias-variance decomposition loss. prove family exact bias-variance-diversity decompositions, losses both regression and classification, e.g., squared, cross-entropy, Poisson losses. For where additive not available (e.g.,...
Understanding body malodour in a measurable manner is essential for developing personal care products. Body the result of bodily secretion highly complex mixture volatile organic compounds. Current measurement methods are manual, time consuming and costly, requiring an expert panel assessors to assign score each human test subject. This article proposes technology-based solution automate this task by custom-designed classification system comprising electronic nose sensor array, readout...
What is the simplest thing you can do to solve a problem? In context of semi-supervised feature selection, we tackle exactly this-how much gain from two simple classifier-independent strategies. If have some binary labelled data and unlabelled, could assume unlabelled are all positives, or them negatives. These minimalist, seemingly naive, approaches not previously been studied in depth. However, with theoretical empirical studies, show they provide powerful results for via hypothesis...
In computer systems, information leaks from the physical hardware through side-channel signals such as power draw. We can exploit these to infer state of ongoing computational tasks without having direct access device. This paper investigates application recent deep learning techniques analysis in both classification machine and anomaly detection. use real data collected three different devices: an Arduino, a Raspberry Pi, Siemens PLC. For we compare performance Multi-Layer Perceptron Long...
Java program execution times vary greatly with different garbage collection algorithms. Until now, it has not been possible to determine the best GC algorithm for aparticular without exhaustively profiling that all available This paper presents a new approach. We use machine learning techniques build prediction model that, given asingle profile run of previously unseen program,can predict good program. implement this technique in Jikes RVM and test onseveral standard benchmark suites. Our...