Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science 2nd Edition Franco Taroni instant download
Bayesian Networks for Probabilistic Inference and Decision Analysis in Forensic Science 2nd Edition Franco Taroni instant download
https://ebookultra.com/download/bayesian-networks-for-probabilistic-
inference-and-decision-analysis-in-forensic-science-2nd-edition-franco-
taroni/
https://ebookultra.com/download/bayesian-networks-and-probabilistic-
inference-in-forensic-science-1st-edition-franco-taroni/
https://ebookultra.com/download/bayesian-analysis-in-natural-language-
processing-2nd-edition-shay-cohen/
https://ebookultra.com/download/data-analysis-a-bayesian-tutorial-2nd-
edition-devinderjit-sivia/
https://ebookultra.com/download/modeling-and-reasoning-with-bayesian-
networks-1st-edition-adnan-darwiche/
STATISTICS IN PRACTICE
Series Advisors
Founding Editor
Vic Barnett
Nottingham Trent University, UK
A complete list of titles in this series appears at the end of the volume.
Trim size: 170mm x 244mm Taroni ffirs.tex V3 - 06/19/2014 10:37 A.M. Page iii
Franco Taroni
University of Lausanne, Switzerland
Alex Biedermann
University of Lausanne, Switzerland
Silvia Bozza
University Ca’ Foscari of Venice, Italy
Paolo Garbolino
University IUAV of Venice, Italy
Colin Aitken
University of Edinburgh, UK
Trim size: 170mm x 244mm Taroni ffirs.tex V3 - 06/19/2014 10:37 A.M. Page iv
The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs
and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form
or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright,
Designs and Patents Act 1988, without the prior permission of the publisher.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in
electronic books.
Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product
names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The
publisher is not associated with any product or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book,
they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and
specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the under-
standing that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be
liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent
professional should be sought.
A catalogue record for this book is available from the British Library.
ISBN: 978-0-470-97973-0
To our families
Trim size: 170mm x 244mm Taroni ffirs.tex V3 - 06/19/2014 10:37 A.M. Page vi
Bayes Nets and Influence Diagrams [...] will be the core elements of new computer
technologies that will make the 21st century the Century of Bayes. [...] Bayes Nets and
Influence Diagrams are not just important design tools; they also represent a major
enhancement of the understanding about how important intellectual tasks typically
performed by people should and can be performed.
– Ward Edwards
Contents
Foreword xiii
viii CONTENTS
CONTENTS ix
x CONTENTS
CONTENTS xi
10 Pre-assessment 314
10.1 Introduction 314
10.2 General elements of pre-assessment 315
10.3 Pre-assessment in a fibre case: A worked through example 316
10.3.1 Preliminaries 316
10.3.2 Propositions and relevant events 317
10.3.3 Expected likelihood ratios 319
10.3.4 Construction of a Bayesian network 321
10.4 Pre-assessment in a cross-transfer scenario 321
10.4.1 Bidirectional transfer 321
10.4.2 A Bayesian network for a pre-assessment of a cross-transfer
scenario 324
10.4.3 The value of the findings 325
10.5 Pre-assessment for consignment inspection 328
10.5.1 Inspecting small consignments 328
10.5.2 Bayesian network for inference about small consignments 330
10.5.3 Pre-assessment for inspection of small consignments 333
10.6 Pre-assessment for gunshot residue particles 335
10.6.1 Formation and deposition of gunshot residue particles 335
10.6.2 Bayesian network for grouped expected findings (GSR counts) 336
10.6.3 Examples for GSR count pre-assessment using a Bayesian
network 339
Trim size: 170mm x 244mm Taroni ftoc.tex V1 - 06/23/2014 12:13 A.M. Page xii
xii CONTENTS
References 419
Foreword
I became a forensic scientist in 1966, training as a document examiner at the Home Office
Forensic Science Laboratory, Cardiff. I recall that, after I had been in the job for 2 or 3 years,
one of my colleagues travelled up to London for a Friday evening meeting of the British
Academy of Forensic Science: a visit to London was something of an adventure in those days
and attendance at such a meeting was regarded as a perk to be savoured. On the following
Monday, I asked my colleague how the meeting had gone and he replied that it was not bad
as meetings went, but it would not have interested me very much. However, he did give me a
photocopy that had been used as a handout at the meeting. My colleague was right in the sense
that I had not paid much attention to the theory of evidence interpretation up to that point, but
he did me an enormous favour because the handout was a prepublication version of a paper
by two Americans called Finkelstein and Fairley1 (1970). This was my first introduction to
Bayesian reasoning, and I must confess that it took me several years to come to terms with it.
Later, having qualified as a statistician, I devoted my efforts largely to the interpretation
of glass refractive index measurements – addressing what we would now call source-level
propositions. This work was entirely frequentist, but again I was fortunate when I learned of
a new paper by Dennis Lindley (1975). This, of course, was also Bayesian and what made
it particularly exciting for me was that it addressed what we would now call offence-level
propositions. In time, I took Dennis’ analysis and modified it to address what we would now
call activity-level propositions, though my first attempt to publish the results gained scathing
criticism from a reviewer and rejection by the journal! The analysis took account of the phe-
nomena of transfer and persistence and also the background distribution of extraneous material
on clothing. At a conference in Sacramento in 1982 and later in Vancouver in 1987, I heard
David Stoney’s talk about the notion of relevance, and I realized that here was a further factor
to be built in to our interpretation. My work in this area culminated in a paper that was pub-
lished in the Journal of the Royal Statistical Society (Evett 1984). Although I was comfortable
with the theory, I realized that it was far too mathematical for practical use as it stood so it
was a great delight for me when, while I was visiting at Victoria University, Wellington, Tony
Vignaux and Bernard Robertson introduced me to Bayes’ nets (Tony and Bernard published a
paper on Bayes’ nets back in 1993). The issues that I had been trying to address by means of
complicated and relatively inaccessible algebra could be captured simply by diagrams such as
1 Many years later, I met both Michael Finkelstein and Will Fairley at a conference in Arizona – it was a great
xiv FOREWORD
those in Figure 6.4 that follows. This was, for me, greatly exciting and the prospects seemed
almost limitless.
During the 1990s, much of my work was with DNA statistics, but I was enormously for-
tunate to be invited to be part of a project called ‘Case Assessment and Interpretation’ (CAI).
This was devoted to pursuing the notion of delivering forensic science in a manner that best
fulfilled the needs of the criminal justice system (CJS). So much came out of this project
that clarified our view of the forensic scientist, and I recall still today the meeting at FSS
Headquarters in Birmingham at which my colleague and friend, Graham Jackson, first used
the phrase ‘hierarchy of propositions’. Using the notion of the hierarchy, we ranged across
forensic science disciplines looking at issues such as the combination of evidence types, the
two-trace problem, and transfer from scene to offender. On reflection, we did not employ
Bayes’ nets nearly as much as we could have done in the early days. However, we were later
joined by a specialist in that field, Roberto Puch-Solis, and he was able to develop software
for implementing several of the CAI initiatives within a Bayes’ nets framework.
The work of the authors of this book shows with clarity how Bayes’ nets clearly establish
inferential structure, for example: Figures 3.1 to 3.4 for the two-trace problem; Figure 5.8
for cross-transfer of evidence; and Figure 6.10 for raising inference up from source-level to
crime-level propositions. One of the important new developments from the CAI project was
the notion of carrying out a pre-assessment of a case to establish the strategy that has the
best expectation of meeting the needs of the CJS. Chapter 10 demonstrates how a Bayes’ nets
approach clarifies thinking and provides structure to this process: Figure 10.1 shows how the
essential structure may be graphically displayed.
It might be an exaggeration to claim that DNA profiling has led to a revolution in forensic
science. Whether or not that is the case, it is inarguable that there have been many new inferen-
tial challenges over the last 20 years. Here too, it is possible to become submerged in complex
mathematics, but again, Bayes’ nets enable the core issues to be exposed diagrammatically.
For example, dealing with what is sometimes called the ‘brother’s defence’ is something that
I handled algebraically back in the 1990s, but here the authors present a more transparent
analysis by means of the Bayes’ net in Figure 7.7. A major area for debate, research and
development is the interpretation of DNA mixtures, again providing challenges that are pecu-
liar to the field. Much elegant work has been done by a number of different groups, and an
introduction is given in Section 7.8, with Figure 7.14 showing how a Bayes’ net captures the
issues of a two-person mixture. Also because of my work on DNA interpretation, I found
myself drawn into the fascinating world of inference of kinship (‘paternity testing’ being just
one aspect of this complex pursuit). It is amazing how complex a kinship analysis can become
in, for example, a case of identification of human remains where comparison samples come
from a range of different possible relatives of the deceased. Later, in 2002, I was fortunate to
be present at the Fifth International Conference on Forensic Inference and Statistics in Venice
when Robert Cowell demonstrated the power of Bayes’ nets for kinship analyses: Figure 7.15
shows how a simple paternity analysis may be represented in this way.
And what of the future? A naïve view of forensic inference that held some sway in the late
twentieth century was that statistical methods, allied with the power of modern computers,
would replace human judgement in due time. However, we have gradually come to see that
the knowledge, judgement and understanding of the forensic practitioner will always be at
the core of evidence interpretation. The notion of the ‘real LR’ is an illusion – any calcula-
tion invokes a model and the validity and parameters of that model rest on the knowledge of
the practitioner who provides the court with an LR that reflects his/her personal judgement.
Trim size: 170mm x 244mm Taroni fbetw.tex V2 - 06/13/2014 9:17 A.M. Page xv
FOREWORD xv
However, no one would wish to see human judgement to be unconstrained, and it is essential
that we have systems in place to calibrate personal probabilities and LRs. There is much good
work currently being carried out in the field of calibration of forensic opinions and, it seems to
me, that work naturally complements the structured approach to logical inference that flows
from the application of Bayes’ nets. Bayesian reasoning provides the logical framework for
bringing together all of the personal probabilities necessary for a given inferential problem:
it is the calibration process that considers the robustness of those probabilities and the Bayes’
net that enables the exploration of the sensitivity of the output, given the calibration data. Pure
synergy: this, for me, is the future of forensic science.
Finally, I wish to compliment the authors on the painstaking scholarly work that has led
to the creation of this book, which should, I believe, have a place on the bookshelf of every
forensic scientist who cares about the science of evidence interpretation.
Ian Evett,
January 2014
Trim size: 170mm x 244mm Taroni fbetw.tex V2 - 06/13/2014 9:17 A.M. Page xvi
Trim size: 170mm x 244mm Taroni fpref01.tex V2 - 06/23/2014 12:18 A.M. Page xvii
Suppose that you are a forensic scientist, facing a large quantity of information coming from
various observations, data or, more generally, findings related to a case under investigation.
Your task is to help express a probabilistic conclusion on the joint value of such a quantity
items of information or to assist a court of justice in expressing a belief on a judicial question
of interest, typically expressed in terms of a proposition, compared to a particular alternative.
How should you proceed? Ten years ago, Professor Dennis Lindley wrote in his foreword for
another book of two of us (Aitken and Taroni 2004, p. 24):
A problem that arises in a courtroom, affecting both lawyers, witnesses and jurors,
is that several pieces of evidence have to be put together before a reasoned judge-
ment can be reached: as when motive has to be considered along with material
evidence. Probability is designed to effect such combinations but the accumula-
tion of simple rules can produce complicated procedures. Methods of handling
sets of evidence have been developed: for example Bayes nets (...). There is a fas-
cinating interplay here between the lawyer and the scientist where they can learn
from each other and develop tools that significantly assist in the production of a
better judicial system.
Indeed, during the past three decades, the so-called Bayesian networks have gradually
become a centre of attention for researchers from several academic fields. Whenever compli-
cated inference problems involving uncertainty as a characterizing feature need to be captured
and approached in a coherent way, that is using the normative framework of probability, their
clarity of formulation and thorough computational architecture can provide a level of assis-
tance that in many fields is unprecedented, in particular when there is a need to associate a
reasoning process with a wider context of decision analysis and decision making.
As pointed out in forensic science and judicial literature, the merit of the Bayesian net-
work graphical probability environment goes well beyond a purely descriptive account that
focusses on the translation of a reasoner’s view of a particular inference problem of interest.
On the one hand, Bayesian networks support the concise description of challenging practical
problems and the communication of their essential features so as to favour their understanding
amongst discussants. On the other hand, Bayesian networks extend to a dynamic dimension
that provides a means for belief computations; that is, the revision of a reasoner’s belief struc-
ture as a result of knowing the truth or otherwise of one or more propositions that are part of
the description of the overall problem. One of the very strengths of Bayesian networks is that
Trim size: 170mm x 244mm Taroni fpref01.tex V2 - 06/23/2014 12:18 A.M. Page xviii
their users can concentrate their efforts on eliciting sensible network structures and probabil-
ity assignments, while leaving the computational burden to computerized implementations of
Bayesian network models. However, there is no claim here of a ‘true’ model: indeed, different
analysts may come up with different models for the same problem. Definitions of basic enti-
ties, the specification of their relationships and probability assignments may naturally differ
because different analysts may hold different background information and may have different
views of how a particular problem ought to be understood. However, this is not a drawback
of the Bayesian network modelling language; it is one of its very strengths to make such dif-
ferences explicit and provide a transparent framework for exploring the nature and extent of
these differences. With respect to the theory of Bayesian networks, this amounts to applied
research, but with respect to forensic science, such research is fundamental because it can
provide original and innovative insights.
Inference and decision analysis, supported by Bayesian networks, should help us to
acquire a better understanding of the problem we face, in terms of the target propositions of
interest, our uncertainties about their true state and the way in which new items of information
ought to affect our view. This better understanding can help to place scientists in a more
secure position when they are required to advise other participants in the legal process on
issues concerning the evaluation of forensic results. Typical questions include, but are not
limited to: What is the bearing this finding has on this proposition, as compared to a given
alternative proposition? If so, to what extent can we affirm degree of support? Should we
attempt to acquire further information? If so, which other information?
One point that is clear from these introductory thoughts is that there are no pre-defined
solutions. Bayesian networks are an abstract concept, and besides some aspects of definition
that prescribe particular modelling constraints, there is nothing in the concept as such to tell
us how to define sensible Bayesian network structures. This places forensic scientists in a
responsible position: they need to make up their minds seriously and invoke further argument
to justify particular model structures and their relevance for particular contexts of application.
Like probability, Bayesian networks are both a very strict and a very liberal concept. They
require the analyst to observe a few general principles of probabilistic reasoning, but beyond
this, there are no prescriptions of as to how the basic terms ought to be interpreted. This
highlights the personal nature of the approach, for which Bayesian analysis in general is so
well known.
At the same time, this paradigm leads directly to one of the main motivations for a book on
Bayesian networks for forensic science. It is driven by the question of how forensic scientists
may use Bayesian networks meaningfully in their work. The idea thus is to offer the reader a
guided introduction to the use of Bayesian networks for analysing forensic inference problems
that arise in connection with various types of traces. To convince the reader that Bayesian
networks can be specified in a defensible way, it is useful to point out that they can capture
and illustrate the rationale behind particular probabilistic solutions, notably likelihood ratio
formulae described in existing literature, which are now generally accepted as a measure of
probative value. This is illustrated through various examples given throughout this book with
reference to the original literature. The aim is to clarify the logic of generic structures of
inferential networks that readers may transfer to their own contexts of application. Often,
original literature provides numerical examples based on scenarios inspired by real cases that
will allow the reader to track particular numerical output (Evett et al. 1998b).
Descriptions and analyses of entire real cases demand a substantial amount of additional
discussion and explanation, in particular with respect to numerical specification. This would
Random documents with unrelated
content Scribd suggests to you:
The Project Gutenberg eBook of Essais de
Montaigne (self-édition) - Volume IV
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.
Translator: Michaud
Language: French
ESSAIS DE MONTAIGNE
Exemplaire
No 127
ESSAIS DE MONTAIGNE
Planche IV
NOTE DE L’ÉDITEUR
Le Général Michaud étant décédé au cours de l’impression du
présent ouvrage, ce IVe volume a été rédigé d’après le texte et les
notes laissées par l’auteur.
FASCICULE A
NOTICE
SUR MONTAIGNE, LES ESSAIS
RÉSUMÉ CHRONOLOGIQUE
DES FAITS PRINCIPAUX DE LA VIE DE MONTAIGNE.
1572.—Massacre de la Saint-Barthélemy.
1574.—Mort de Charles IX, avènement de Henri III.
1587.—Bataille de Coutras.
SA VIE.
[4] La Cour des aides était une chambre jugeant en dernier ressort les
questions afférentes aux aides, subsides établis jadis sur les boissons pour
subvenir aux dépenses de l’Etat; et ultérieurement et par extension tous
autres impôts.
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
ebookultra.com