QG/GA: a stochastic search for Progol

Most search techniques within ILP require the evaluation of a large number of inconsistent clauses. However, acceptable clauses typically need to be consistent, and are only found at the “fringe” of the search space. A search approach is presented, based on a novel algorithm called QG (Quick Generalization). QG carries out a random-restart stochastic bottom-up search which efficiently generates a consistent clause on the fringe of the refinement graph search without needing to explore the graph in detail. We use a Genetic Algorithm (GA) to evolve and re-combine clauses generated by QG. In this QG/GA setting, QG is used to seed a population of clauses processed by the GA. Experiments with QG/GA indicate that this approach can be more efficient than standard refinement-graph searches, while generating similar or better solutions. Editors: Ramon Otero, Simon Colton.     

ILP turns 20

Inductive Logic Programming (ILP) is an area of Machine Learning which has now reached its twentieth year. Using the analogy of a human biography this paper recalls the development of the subject from its infancy through childhood and teenage years. We show how in each phase ILP has been characterised by an attempt to extend theory and implementations in tandem with the development of novel and challenging real-world applications. Lastly, by projection we suggest directions for research which will help the subject coming of age.     

Beneficial and harmful explanatory machine learning

Machine Learning - Given the recent successes of Deep Learning in AI there has been increased interest in the role and need for explanations in machine learned theories. A distinct notion in this...     

Protein secondary structure prediction using logic-based machine learning

Protein Engineering, 5, 647–657 An error was found (Leberman, 1993) in the selection of non-homologousproteins used to train the machine learning program Golem. Theproteins 155C (cytochrome C550) and 2C2C (cytochrome C2 oxidized)have 43.5% sequence homology. Removal of the result for 2C2C(the most accurately predicted of the two proteins) producesa Q3 accuracy of 78% (no change) and a Matthew's correlationof 0.53 (a reduction of 0.04). Removal of 2C2C makes no differenceto results for the independent test set of non-homologous proteins. Reference. Leberman, R. (1993) Protein Engng, 6, 547.     

The effect of holding adult Oryzaephilus surinamensis (L.) (Coleoptera: Silvanidae) at different temperatures on their subsequent susceptibility to malathion

There is reason to believe that the temperature prior to exposure may effect the response of adult Oryzaephilus surinamensis to the recommended discriminating dose test for malathion resistance. Experiments were carried out in which four malathion resistant strains of O. surinamensis were kept at 15, 20, 25 and 30°C for two weeks prior to testing at 25 C. It was found that conditioning at the lower temperatures resulted in a significantly higher knockdown than conditioning at the higher temperatures. For each 5°C fall in conditioning temperature there was a 7.5% increase in knockdown. Chemical analysis showed that in the first hour of exposure more malathion was taken up by the beetles conditioned at lower temperatures than by those conditioned at the higher temperatures.     

Learning higher-order logic programs

Machine Learning - A key feature of inductive logic programming is its ability to learn first-order programs, which are intrinsically more expressive than propositional programs. In this paper, we...     

Mapping the Underworld – State-of-the-art review

A major UK initiative, entitled Mapping the Underworld (MTU), is seeking to address the serious social, environmental and economic consequences arising from an inability to locate accurately and completely the buried utility service infrastructure without resorting to excavations. One of the four MTU projects aims to develop and prove the efficacy of a multi-sensor device for accurate remote buried utility service detection, location and, where possible, utility identification. This paper aims to introduce the MTU programme followed by a state-of-the-art review of the three essential technologies that are to be combined in the device – ground penetrating radar (GPR), low-frequency quasi-static electromagnetic fields and acoustics – and a summary of the influence of different soil types and states on the transmission of the various signals, and therefore how the techniques might be optimised from a knowledge of the ground instead of using very broad simplifying assumptions. The latest developments in impulse GPR, frequency modulated continuous waveform (FMCW) GPR and stepped frequency continuous waveform (SFCW) GPR are described and previous attempts to combine GPR with other sensing technologies are introduced. The work on quasi-static fields explores the ‘fields-of-opportunity’ related to the 50 Hz currents flowing in existing underground power circuits and the electric field variations when low-frequency current in actively induced into the ground. Acoustic techniques have been primarily used for leak detection and the review focuses on the potential for their application to buried utility service location. The paper concludes with a discussion of the facilities required, and currently available, for comprehensive assessment and independent verification of the performance of both existing devices/technologies and of the multi-sensor device under development.