Richard J. Preen

Jim Smith, Richard J. Preen, Andrew McCarthy et al.

We present SACRO-ML, an integrated suite of open source Python tools to facilitate the statistical disclosure control (SDC) of machine learning (ML) models trained on confidential data prior to public release. SACRO-ML combines (i) a SafeModel package that extends commonly used ML models to provide ante-hoc SDC by assessing the vulnerability of disclosure posed by the training regime; and (ii) an Attacks package that provides post-hoc SDC by rigorously assessing the empirical disclosure risk of a model through a variety of simulated attacks after training. The SACRO-ML code and documentation are available under an MIT license at https://github.com/AI-SDC/SACRO-ML

Emily Jefferson, James Liley, Maeve Malone et al.

TREs are widely, and increasingly used to support statistical analysis of sensitive data across a range of sectors (e.g., health, police, tax and education) as they enable secure and transparent research whilst protecting data confidentiality. There is an increasing desire from academia and industry to train AI models in TREs. The field of AI is developing quickly with applications including spotting human errors, streamlining processes, task automation and decision support. These complex AI models require more information to describe and reproduce, increasing the possibility that sensitive personal data can be inferred from such descriptions. TREs do not have mature processes and controls against these risks. This is a complex topic, and it is unreasonable to expect all TREs to be aware of all risks or that TRE researchers have addressed these risks in AI-specific training. GRAIMATTER has developed a draft set of usable recommendations for TREs to guard against the additional risks when disclosing trained AI models from TREs. The development of these recommendations has been funded by the GRAIMATTER UKRI DARE UK sprint research project. This version of our recommendations was published at the end of the project in September 2022. During the course of the project, we have identified many areas for future investigations to expand and test these recommendations in practice. Therefore, we expect that this document will evolve over time.

Richard J. Preen, Jim Smith

Machine learning models can inadvertently expose confidential properties of their training data, making them vulnerable to membership inference attacks (MIA). While numerous evaluation methods exist, many require computationally expensive processes, such as training multiple shadow models. This article presents two new complementary approaches for efficiently identifying vulnerable tree-based models: an ante-hoc analysis of hyperparameter choices and a post-hoc examination of trained model structure. While these new methods cannot certify whether a model is safe from MIA, they provide practitioners with a means to significantly reduce the number of models that need to undergo expensive MIA assessment through a hierarchical filtering approach. More specifically, it is shown that the rank order of disclosure risk for different hyperparameter combinations remains consistent across datasets, enabling the development of simple, human-interpretable rules for identifying relatively high-risk models before training. While this ante-hoc analysis cannot determine absolute safety since this also depends on the specific dataset, it allows the elimination of unnecessarily risky configurations during hyperparameter tuning. Additionally, computationally inexpensive structural metrics serve as indicators of MIA vulnerability, providing a second filtering stage to identify risky models after training but before conducting expensive attacks. Empirical results show that hyperparameter-based risk prediction rules can achieve high accuracy in predicting the most at risk combinations of hyperparameters across different tree-based model types, while requiring no model training. Moreover, target model accuracy is not seen to correlate with privacy risk, suggesting opportunities to optimise model configurations for both performance and privacy.

Richard J. Preen, Larry Bull

This article presents the first results from using a learning classifier system capable of performing adaptive computation with deep neural networks. Individual classifiers within the population are composed of two neural networks. The first acts as a gating or guarding component, which enables the conditional computation of an associated deep neural network on a per instance basis. Self-adaptive mutation is applied upon reproduction and prediction networks are refined with stochastic gradient descent during lifetime learning. The use of fully-connected and convolutional layers are evaluated on handwritten digit recognition tasks where evolution adapts (i) the gradient descent learning rate applied to each layer (ii) the number of units within each layer, i.e., the number of fully-connected neurons and the number of convolutional kernel filters (iii) the connectivity of each layer, i.e., whether each weight is active (iv) the weight magnitudes, enabling escape from local optima. The system automatically reduces the number of weights and units while maintaining performance after achieving a maximum prediction error.

Richard J. Preen, Stewart W. Wilson, Larry Bull

Autoencoders are data-specific compression algorithms learned automatically from examples. The predominant approach has been to construct single large global models that cover the domain. However, training and evaluating models of increasing size comes at the price of additional time and computational cost. Conditional computation, sparsity, and model pruning techniques can reduce these costs while maintaining performance. Learning classifier systems (LCS) are a framework for adaptively subdividing input spaces into an ensemble of simpler local approximations that together cover the domain. LCS perform conditional computation through the use of a population of individual gating/guarding components, each associated with a local approximation. This article explores the use of an LCS to adaptively decompose the input domain into a collection of small autoencoders where local solutions of different complexity may emerge. In addition to benefits in convergence time and computational cost, it is shown possible to reduce code size as well as the resulting decoder computational cost when compared with the global model equivalent.

Richard J. Preen, Larry Bull, Andrew Adamatzky

The use of high-fidelity computational simulations promises to enable high-throughput hypothesis testing and optimisation of cancer therapies. However, increasing realism comes at the cost of increasing computational requirements. This article explores the use of surrogate-assisted evolutionary algorithms to optimise the targeted delivery of a therapeutic compound to cancerous tumour cells with the multicellular simulator, PhysiCell. The use of both Gaussian process models and multi-layer perceptron neural network surrogate models are investigated. We find that evolutionary algorithms are able to effectively explore the parameter space of biophysical properties within the agent-based simulations, minimising the resulting number of cancerous cells after a period of simulated treatment. Both model-assisted algorithms are found to outperform a standard evolutionary algorithm, demonstrating their ability to perform a more effective search within the very small evaluation budget. This represents the first use of efficient evolutionary algorithms within a high-throughput multicellular computing approach to find therapeutic design optima that maximise tumour regression.

Richard J. Preen, Jiseon You, Larry Bull et al.

Microbial fuel cells (MFCs) perform wastewater treatment and electricity production through the conversion of organic matter using microorganisms. For practical applications, it has been suggested that greater efficiency can be achieved by arranging multiple MFC units into physical stacks in a cascade with feedstock flowing sequentially between units. In this paper, we investigate the use of computational intelligence to physically explore and optimise (potentially) heterogeneous MFC designs in a cascade, i.e. without simulation. Conductive structures are 3-D printed and inserted into the anodic chamber of each MFC unit, augmenting a carbon fibre veil anode and affecting the hydrodynamics, including the feedstock volume and hydraulic retention time, as well as providing unique habitats for microbial colonisation. We show that it is possible to use design mining to identify new conductive inserts that increase both the cascade power output and power density.

Richard J. Preen, Larry Bull

An initial study of surrogate-assisted evolutionary algorithms used to design vertical-axis wind turbines wherein candidate prototypes are evaluated under fan generated wind conditions after being physically instantiated by a 3D printer has recently been presented. Unlike other approaches, such as computational fluid dynamics simulations, no mathematical formulations were used and no model assumptions were made. This paper extends that work by exploring alternative surrogate modelling and evolutionary techniques. The accuracy of various modelling algorithms used to estimate the fitness of evaluated individuals from the initial experiments is compared. The effect of temporally windowing surrogate model training samples is explored. A surrogate-assisted approach based on an enhanced local search is introduced; and alternative coevolution collaboration schemes are examined.

Richard J. Preen, Larry Bull

The production of renewable and sustainable energy is one of the most important challenges currently facing mankind. Wind has made an increasing contribution to the world's energy supply mix, but still remains a long way from reaching its full potential. In this paper, we investigate the use of artificial evolution to design vertical-axis wind turbine prototypes that are physically instantiated and evaluated under fan generated wind conditions. Initially a conventional evolutionary algorithm is used to explore the design space of a single wind turbine and later a cooperative coevolutionary algorithm is used to explore the design space of an array of wind turbines. Artificial neural networks are used throughout as surrogate models to assist learning and found to reduce the number of fabrications required to reach a higher aerodynamic efficiency. Unlike in other approaches, such as computational fluid dynamics simulations, no mathematical formulations are used and no model assumptions are made.

Richard J. Preen, Larry Bull

We have recently presented an initial study of evolutionary algorithms used to design vertical-axis wind turbines (VAWTs) wherein candidate prototypes are evaluated under approximated wind tunnel conditions after being physically instantiated by a 3D printer. That is, unlike other approaches such as computational fluid dynamics simulations, no mathematical formulations are used and no model assumptions are made. However, the representation used significantly restricted the range of morphologies explored. In this paper, we present initial explorations into the use of a simple generative encoding, known as Gielis superformula, that produces a highly flexible 3D shape representation to design VAWT. First, the target-based evolution of 3D artefacts is investigated and subsequently initial design experiments are performed wherein each VAWT candidate is physically instantiated and evaluated under approximated wind tunnel conditions. It is shown possible to produce very closely matching designs of a number of 3D objects through the evolution of supershapes produced by Gielis superformula. Moreover, it is shown possible to use artificial physical evolution to identify novel and increasingly efficient supershape VAWT designs.