Learning and using taxonomies for visual and olfactory classification

Humans are able of distinguishing more than 5000 visual categories even in complex environments using a variety of different visual systems all working in tandem. We seem to be capable of distinguishing thousands of different odors as well. In the machine learning community, many commonly used multi-class classifiers do not scale well to such large numbers of categories. This thesis demonstrates a method of automatically creating application-specific taxonomies to aid in scaling classification algorithms to more than 100 cate- gories using both visual and olfactory data. The visual data consists of images collected online and pollen slides scanned under a microscope. The olfactory data was acquired by constructing a small portable sniffing apparatus which draws air over 10 carbon black polymer composite sensors. We investigate performance when classifying 256 visual categories, 8 or more species of pollen and 130 olfactory categories sampled from common household items and a standardized scratch-and-sniff test. Taxonomies are employed in a divide-and-conquer classification framework which improves classification time while allowing the end user to trade performance for specificity as needed. Before classification can even take place, the pollen counter and electronic nose must filter out a high volume of background “clutter” to detect the categories of interest. In the case of pollen this is done with an efficient cascade of classifiers that rule out most non-pollen before invoking slower multi-class classifiers. In the case of the electronic nose, much of the extraneous noise encountered in outdoor environments can be filtered using a sniffing strategy which preferentially samples the visensor response at frequencies that are relatively immune to background contributions from ambient water vapor. This combination of efficient background rejection with scalable classification algorithms is tested in detail for three separate projects: 1) the Caltech-256 Image Dataset, 2) the Caltech Automated Pollen Identification and Counting System (CAPICS) and 3) a portable electronic nose specially constructed for outdoor use.

Cryptanalysis of the convex hull click human identification protocol

Recently, a convex hull-based human identification protocol was proposed by Sobrado and Birget, whose steps can be performed by humans without additional aid. The main part of the protocol involves the user mentally forming a convex hull of secret icons in a set of graphical icons and then clicking randomly within this convex hull. While some rudimentary security issues of this protocol have been discussed, a comprehensive security analysis has been lacking. In this paper, we analyze the security of this convex hull-based protocol. In particular, we show two probabilistic attacks that reveal the user’s secret after the observation of only a handful of authentication sessions. These attacks can be efficiently implemented as their time and space complexities are considerably less than brute force attack. We show that while the first attack can be mitigated through appropriately chosen values of system parameters, the second attack succeeds with a non-negligible probability even with large system parameter values that cross the threshold of usability.

Cryptanalysis of the convex hull click human identification protocol

Recently a convex hull based human identification protocol was proposed by Sobrado and Birget, whose steps can be performed by humans without additional aid. The main part of the protocol involves the user mentally forming a convex hull of secret icons in a set of graphical icons and then clicking randomly within this convex hull. In this paper we show two efficient probabilistic attacks on this protocol which reveal the user’s secret after the observation of only a handful of authentication sessions. We show that while the first attack can be mitigated through appropriately chosen values of system parameters, the second attack succeeds with a non-negligible probability even with large system parameter values which cross the threshold of usability.

A new human identification protocol and Coppersmith's baby-step giant-step algorithm

We propose a new protocol providing cryptographically secure authentication to unaided humans against passive adversaries. We also propose a new generic passive attack on human identification protocols. The attack is an application of Coppersmith’s baby-step giant-step algorithm on human identification protcols. Under this attack, the achievable security of some of the best candidates for human identification protocols in the literature is further reduced. We show that our protocol preserves similar usability while achieves better security than these protocols. A comprehensive security analysis is provided which suggests parameters guaranteeing desired levels of security.

Foodborne Yersinia : Identification and Molecular Epidemiology of Isolates from Human Infections

Yersinia enterocolitica and Yersinia pseudotuberculosis are among the major enteropathogenic bacteria causing infections in humans in many industrialized countries. In Finland, Y. pseudotuberculosis has caused 10 outbreaks among humans during 1997-2008. Some of these outbreaks have been very extensive involving over 400 cases; mainly children attending schools and day-care. Y. enterocolitica, on the contrary, has caused mainly a large number of sporadic human infections in Finland. Y. pseudotuberculosis is widespread in nature, causing infections in a variety of domestic and wild animals. Foodborne transmission of human infections has long been suspected, however, attempts to trace the pathogen have been unsuccessful before this study that epidemiologically linked Y. pseudotuberculosis to a specific food item. Furthermore, due to modern food distribution systems, foodborne outbreaks usually involve many geographically separate infection clusters difficult to identify as part of the same outbreak. Among pathogenic Y. enterocolitica, the global predominance of one genetically homogeneous type (bioserotype 4/O:3) is a challenge to the development of genetic typing methods discriminatory enough for epidemiological purposes, for example, for tracing back to the sources of infections. Furthermore, the diagnostics of Y. enterocolitica infections is hampered because clinical laboratories easily misidentify some other members of the Yersinia species (Y. enterocolitica–like species) as Y. enterocolitica. This results in misleading information on the prevalence and clinical significance of various Yersinia isolates. The aim of this study was to develop and optimize molecular typing methods to be used in epidemiological investigations of Y. enterocolitica and Y. pseudotuberculosis, particularly in active surveillance and outbreak investigations of Y. pseudotuberculosis isolates. The aim was also to develop a simplified set of phenotypic tests that could be used in routine diagnostic laboratories for the correct identification of Y. enterocolitica and Y. enterocolitica –like species. A PFGE method designed here for typing of Y. pseudotuberculosis was efficient in linking the geographically dispersed and apparently unrelated Y. pseudotuberculosis infections as parts of the same outbreak. It proved to be useful in active laboratory-based surveillance of Y. pseudotuberculosis outbreaks. Throughout the study period, information about the diversity of genotypes among outbreak and non-outbreak related strains of human origin was obtained. Also, to our knowledge, this was the first study to epidemiologically link a Y. pseudotuberculosis outbreak of human illnesses to a specific food item, iceberg lettuce. A novel epidemiological typing method based on the use of a repeated genomic region (YeO:3RS) as a probe was developed for the detection and differentiation between strains of Y. enterocolitica subspecies palearctica. This method was able to increase the discrimination in a set of 106 previously PFGE typed Finnish Y. enterocolitica bioserotype 4/O:3 strains among which two main PFGE genotypes had prevailed. The developed simplified method was a more reliable tool than the commercially available biochemical test kits for differentiation between Y. enterocolitica and Y. enterocolitica –like species. In Finland, the methods developed for Y. enterocolitica and Y. pseudotuberculosis have been used to improve the identification protocols and in subsequent outbreak investigations.

Fault Attacks on Pairing-Based Protocols Revisited

Several papers have studied fault attacks on computing a pairing value e(P, Q), where P is a public point and Q is a secret point. In this paper, we observe that these attacks are in fact effective only on a small number of pairing-based protocols, and that too only when the protocols are implemented with specific symmetric pairings. We demonstrate the effectiveness of the fault attacks on a public-key encryption scheme, an identity-based encryption scheme, and an oblivious transfer protocol when implemented with a symmetric pairing derived from a supersingular elliptic curve with embedding degree 2.

Automatic identification and enumeration of algae

A good understanding of the population dynamics of algal communities is vital in many ecological and pollution studies of freshwater and oceanic systems. Present methods require manual counting and identification of algae and can take up to 90 min to obtain a statistically reliable count on a complex population. Several alternative techniques to accelerate the process have been tried on marine samples but none have been completely successful because insufficient effort has been put into verifying the technique before field trials. The objective of the present study has been to assess the potential of in vivo fluorescence of algal pigments as a means of automatically identifying algae. For this work total fluorescence spectroscopy was chosen as the observation technique.

旌节花科的分类和系统学研究

旌节花科是东亚特有科，仅有旌节花属1属。本文研究了该科植物的生活习性、外部形态、花粉形态和外壁超微结构以及花器官发生，并根据多个DNA区段的序列探讨了科的系统位置和属下关系。在此基础上，综合各方面的证据，对旌节花科进行了全面的分类学修订，完善了该科的分类系统。主要内容包括： 　　1、习性和外部形态学 　　结合室内标本和野外考察，对旌节花科植物的生活习性、叶、花序、花和果实各个性状的变异式样、变异幅度、变异规律及其相关性进行了深入研究，并讨论了它们的分类学价值。结果表明：叶片形状、叶缘齿、花序下的叶片是否宿存以及花序是否具花序梗等性状在种间变异非常大，单个性状对旌节花属属下分类意义不大;而叶片毛被、脉序和结构，花序长度，花的大小和颜色以及果实大小等性状通常较稳定，可作为旌节花属划分种和变种的重要依据。 　　2、花粉形态学和外壁超微结构 　　在透射电镜下研究了旌节花属5个代表种和1个外类群（Crossosoma californicum）的花粉外壁超微结构，并在扫描电镜下研究了外类群Crossosomataceae 1个代表种（Crossosoma californicum）的花粉形态学特征，结合前人的研究，探讨了上述性状的分类学意义。结果表明：旌节花属植物花粉形态的共同特征是：花粉粒球形或近球形，三孔沟;外壁具穴状纹饰，穴较细小;外壁内壁分层明显，容易区分（柳叶旌节花除外），覆盖层光滑，几无附属物，柱状层明显，穿孔较少，不形成半覆盖层。各种植物花粉外壁超微结构的特征在旌节花属两个组中较一致，但其差异可作为探讨属下种的分类和系统学关系的依据。此外，对旌节花科和Crossosomataceae花粉表面纹饰和外壁超微结构的研究结果提示：这两个科与省沽油科和山茶科厚皮香亚科有较近缘的关系，而不支持这两个科的姐妹群关系。 　　3、花器官发生 　　首次对旌节花属常绿组和落叶组共2个代表种的花器官发生进行了研究，结果表明：该科植物花为功能性单性花，雌雄异株，花发育过程长达10个月，其中2-3个月是冬季休眠期。在花发育早期，雄花和雌花的发育没有差异，花性别的分化发生在花发育后期，即每年春天花开前后的2-3天，雌花中雌蕊的子房开始膨大，柱头伸长，胚珠多数，能育，而雄蕊逐渐干瘪收缩直至最后花粉不育;此时雄花的雄蕊发育良好，花丝粗壮发达，花药饱满，花粉能育，而雌蕊发育逐渐停滞直至败育。花器官发生方式为向心式，发生顺序为：苞片—外轮萼片—内轮萼片—花瓣—外轮雄蕊—内轮雄蕊—雌蕊。花萼4枚，两轮，交互对生，花瓣和雄蕊均4枚一轮几乎同时发生，花瓣覆瓦状排列。旌节花属常绿组和落叶组植物的花器官发生发育过程没有差别。 　　4、基于多个DNA区段序列的系统发育分析 　　首次选取进化速率较快的5个DNA区段：核核糖体ITS区和叶绿体rps16, ndhF，trnL-F及trnS-G-G区段，对旌节花属分别代表常绿组和落叶组的10种2变种及外类群Crossosomataceae和省沽油科3种植物的相关序列进行测定，并将上述序列合并构建系统树。研究结果表明：旌节花科作为一个单系群得到很好的支持。旌节花属落叶组和常绿组的划分在基于ITS、叶绿体4个区段单独分析及联合分析得到的系统树中没有得到支持，表现在该联合分析得到的树的两个分支与该属植物的习性和外部形态特征无关，而与它们的地理分布密切相关：其中一支主要分布在东亚的偏北部，另一支主要分布在东亚的南部地区，常绿和落叶的种类则混合在这两支中。此外，分子序列分析的结果对该属植物的分类、系统关系、起源和该科的系统位置均有重要意义。系统树显示出分布于台湾的S. sigeyosii与分布于日本的早春旌节花聚成稳定的一支，说明前者与后者，而不是与西域旌节花关系密切。叶绿体基因树与核基因树存在冲突，表明该属经历过网状进化。推测变种披针叶旌节花与云南旌节花可能是杂交起源。根据叶绿体rbcL序列得出的系统树显示旌节花科与产北美的Crossosomataceae的姊妹群关系支持率低于50%。基于最大似然树，运用罚分似然法推算出这两个科分化时间为68.25 ± 10.36 MYA，这两者共同与省沽油科近缘。 　　5、分类学研究 　　综合上述研究结果，对旌节花属进行了分类学修订。属下不设立组;确认该属植物有6种6变种，其中中国特有种3个，变种3个。5种5变种被作为新异名，新组合3个，其中2个改级新组合。对西域旌节花、云南旌节花和滇缅旌节花等进行了模式指定。本文还给出了分种和变种检索表、各类群的形态学描述、地理分布图、标本引证和讨论等。 　　

入侵植物紫茎泽兰遗传变异及克隆多样性研究

现存植被与其花粉分布的关系可作为利用化石花粉图谱重建古植被时的参照，因此，理解现存表土花粉分布格局对解释花粉图谱十分关键。提高花粉鉴定精度是孢粉学和古生态学家一直努力想要解决的关键问题。目前的孢粉学方法将表土花粉鉴定到属的水平都十分困难，因此造成依据花粉图谱进行植被重建的时候出现无法矫正的误差。本研究用分子生物学方法将藜科植物表土花粉鉴定精度从科的水平提高到种的水平。表土花粉样品取自新疆中部样带，利用样带内所有出现的藜科植物共19个种建立藜科植物核基因内转录间隔1区（ITS1）序列库，然后通过巢式聚合酶链式反应（PCR）进行单粒表土花粉的ITS1序列扩增和测序，与序列库中的序列进行比对，从而确定单粒花粉来自哪种藜科植物。这种相对简单的以PCR为基础的方法可以将表土花粉鉴定到种，使在种的水平解释植被与花粉分布的关系成为可能。 紫茎泽兰（Eupatorium adenophorum）是广泛分布在中国西南的入侵植物。以无融合生殖为主，其压条性克隆繁殖方式经常被忽视，本文以紫茎泽兰为例，研究无融合生殖和压条性克隆繁殖为主的入侵种群遗传变异与克隆多样性，分析无融合生殖入侵种群具有高度竞争力和适应能力的分子机制。实验用AFLP分子标记检测了包括缅甸一个种群在内的17个种群的基因组多样性，结果表明，紫茎泽兰是多克隆植物，有丰富的基因型数量，遗传多样性水平低（He＝0.0439），大部分（73.59%）遗传变异存在于种群内，种群间存在显著遗传分化。紫茎泽兰克隆生长策略为游击型，倒伏性克隆繁殖有助于小范围内种群的扩张。各种群基因型组成存在很大差异，但基因型间相似性很高，有相当多的基因型可能来自遗传重组。紫茎泽兰的有性繁殖比例可能比以往所推测的高得多，有助于保持基因型的多样性。

Characterization and environmental interpretation of a floodplain Holocene paleosoil: Implications for paleohydrological reconstructions in the Upper Paraná River, Brazil

Alluvial paleosoil horizons from the 3 to 4 m high banks of the Paraná River were used to reconstruct the palaeoenvironmental history and palaeohydrological regime of the river. The bank consists of a Middle to Late Holocene sequence of sandy to silty clay layers of overbank deposits. The paleosoil horizon is located to a depth of 1.95 m from the top of the bank and is recognized as a guide horizon throughout the floodplain of the Upper Paraná River floodplain. Analyses of organic matter (δ13C and humic/fulvic acids), palynology (pollen and charcoal fragments), magnetic susceptibility, micromorphology, x-ray diffraction and 14C dating were conducted for samples from two representative profiles of the study area. Two phases were characterized in the history of the river: 1) An older phase, of stability in fluvial hydrology (stasis) with low frequency of floods, which produced conditions for soil development (14C 1700 ± 70 yr. BP). At this period, the predominance of herbaceous vegetation (determinated by pollen and δ13C analyses) suggests a climate less humid than the present one, and 2) A second phase, when climate changed to the present conditions of humidity (annual rain fall of 1600 rum) and characterized by the predominance of C3 plants. Under this new hydrological regime, the river developed an agradational floodplain, with a depositional sedimentary rate of 1.2 mm.y-1. © 2006 Gebrüder Borntraeger.

Palynologie und Sedimentologie der Interglazialprofile Döttingen, Bonstorf, Munster und Bilshausen

Palynologie und Sedimentologie der Interglazialprofile Döttingen, Bonstorf, Munster und Bilshausen Zusammenfassung In der vorliegenden Dissertation wurden vier dem Holstein-Interglazial zugehörige Bohrkerne sowie ein rhumezeitlicher Bohrkern palynologisch und sedimentologisch bearbeitet. Die holsteinzeitlichen Bohrkerne stammen aus Kieselgurlagerstätten der Lüneburger Heide (Bonstorf und Munster) und aus einem Trockenmaar (Döttingen) in der Eifel. Der rhumezeitliche Kern stammt aus der Typlokalität Bilshausen im Harzvorland. Neben Prozentwertdiagrammen werden Pollendichte- und wenn möglich Polleninfluxwerte vorgestellt, die insbesondere für die Lokalitäten Hetendorf/Bonstorf und Munster/Breloh bisher nicht verfügbar waren. Mit dem Profil Döttingen konnte erstmals eine sowohl vollständige als auch nicht innerhalb des klassischen Aufkommens holsteinzeitlicher Fundstellen im norddeutschen Tiefland positionierte Holsteinsequenz aus dem deutschen Mittelgebirge dokumentiert werden. Das erhaltene Pollendiagramm bestätigt die aus den norddeutschen Profilen bekannte holsteintypische Vegetationsabfolge, durch die das Holstein gegenüber anderen Interglazialen wie Holozän, Eem oder Rhume palynologisch definiert ist. Neben der grundsätzlichen Übereinstimmung der Pollensequenz unterscheidet sich das Profil Döttingen aber deutlich im prozentualen Aufkommen der beteiligten Taxa von den norddeutschen Profilen. So wird eine hohe Alnus-Präsenz als Merkmal deutscher Holsteinprofile bestätigt, jedoch ist die, in den norddeutschen Lokalitäten durchhaltend hohe oder dominante Beteiligung von Pinus im deutschen Mittelgebirge nicht vorhanden und muss daher auf die Standortbedingungen Norddeutschlands zurückgeführt werden. Abies dagegen ist im Holstein der Mittelgebirge wesentlich präsenter als im norddeutschen Flachland. Im Profil Döttingen wurden insgesamt 10 Tephralagen gefunden. Auf eine dieser Tephren folgt ein „Birken-Kiefern-Gräser Vorstoß“, der palynostratigraphisch dem älteren „Birken-Kiefern Vorstoß“ in Munster/Breloh entspricht. Als eine Typologie des Holstein kann das in den Profilen Döttingen und Munster bestätigte intraholsteinzeitliche Carpinus-Minimum verstanden werden. An Hand sedimentologischer und palynologischer Befunde aus dem Bohrkern MU 2 muss die Existenz zweier, in der Literatur postulierter, postholsteinzeitlicher, „Nachschwankungen“ in Munster/Breloh in Frage gestellt, wenn nicht abgelehnt werden. In Kern MU 2 fallen palynostratigraphische Grenzen häufig mit Sandeinschaltungen zusammen. Eine dieser Sandeinschaltungen, nämlich unmittelbar vor dem älteren „Birken-Kiefern-Vorstoß“, korreliert in ihrer stratigraphischen Position mit der den „Birken-Kiefern-Gräser-Vorstoß“ im Profil Döttingen einleitenden Tephralage. Es gelang die Dauer des intraholsteinzeitlichen Carpinus Minimums auf etwa 1500±100 Jahre zu bestimmen und eine interne Zweigliederung zu dokumentieren. Im rhumezeitlichen Kern von Bilshausen (BI 1) konnten zahlreiche Störungen nachgewiesen werden. Insbesondere im Teufenbereich des Bilshausener „Birken-Kiefern-Vorstoßes“ deuten diese auf eine möglicherweise verfälschte Überlieferung. Der palynologisch markante „Lindenfall“ von Bilshausen liegt im Bereich einer isoklinalen Schichtenverfaltung. Die in der Literatur im Horizont des „Lindenfalls“ beschriebene „Bilshausentephra“ wurde nicht gefunden. Warvenzählungen an den Kernen MU 1, MU 2 und BI 1 ermöglichten Pollenzonendauern in Holstein- und Rhume-Interglazial zu bestimmen. Dabei wurde mittels den Warvenzählungen, unter zu Hilfenahme von Literaturdaten und von Schätzwerten eine Dauer für das Holstein-Interglazial sensu stricto (Pollenzonen I-XIV) von 15400-17800 Jahren und für das Rhume-Interglazial von wahrscheinlich 22000 Jahren bis maximal 26000 Jahren ermittelt.