Philosophiae Doctor - PhD / This study sought to examine independent clause Sesotho personal names as authentic social discourse using the Systemic Functional Linguistics (SFL) theory. It sought to analyze their structure and map them to social functions to demonstrate that they are enacted messages in socio-cultural context of Basotho. It used a form-meaning approach to interpret Sesotho names in socio-cultural contexts of use (cf. Halliday 1994, 2001, Eggins, 1996, 2004 and Martin & Rose 2007) as an alternative to the current formalist approach to onomastica interpretation. The SFL analysis was compared and contrasted mainly with the formalist syntactic specific and semantic specific analyses currently in use by Guma, Sesotho Academy and subsequent authors of Sesotho grammar and other linguists. The purpose of displaying these names as texts in social context enfolded the intent to reflect a systemic interface of lexico-grammar and social activity. The study used the clause-text-culture paradigm to explore Sesotho names as texts or semantic units. The idea was to access their ‘meanings beyond the clause’ (Martin & Rose 2007). Data was collected from national examinations pass lists, admission and employment roll lists from Public, Private, Tertiary, Orphanage institutions. Other data was identified in Telephone directories and Media. The purely linguistic lexico-grammatic analysis of the structure of names was supplemented by interview data from real interpretations from families, owners and senior citizens who have social and cultural knowledge of the meanings of some names. The study has established that Sesotho personal names can present as an independent clause feature. Sesotho personal names can also be described as lexico-grammatical properties and are meaningful in social contexts. They are used to exchange information as statements, demands and commands, and as questions and as exclamations. This means that these names can be categorized according to Halliday’s Mood types which make them function as declaratives, imperatives, interrogatives and exclamatives depending on the awarder’s evaluation. The study also finds that in negotiating attitudes, modality is highly incorporated. The study concludes that Sesotho names conform to the logical structures of the nominal group and the verbal group and these groups reciprocate in use. The verbal group is the core constituent in these names and it serves as a foundation for the nominal and verbal groups particularly because they function as reciprocating propositions. This includes the names with the sub-modification features. This extends the formalist description of Sesotho independent clause in that the identified sub-modifications which are opague and taken for granted by formalist analysts of Sesotho, are explicated as essential elements embedded in the formmeaning relation in SFL. The main contribution is that this is the only study on SFL and onomastica. There is no study that has been conducted using SFL to describe African names. It presents that Sesotho personal names are texts that have been negotiated in socio-cultural contexts. It provides a major departure from most studies that have used the Chomskian formulations or other sociolinguistic theories to describe the naming systems. It displays the art and importance of language use based on experience and culture in the naming system. The study also contributes to fields such as education, history, and others. Lastly, the study has established a new relation of onomastica and SFL theory and onomastica can now be added to the areas “being recognized as providing a very useful descriptive and interpretive framework for viewing language as a strategic, meaning-making resource.” (Eggins 1996:1).

Herpes simplex virus type 1 (HSV-1) has a large linear double-stranded DNA genome in an icosahedral capsid shell, a cell-derived lipid envelope and a proteinaceous tegument layer. There are over fifty viral proteins and many host proteins identified in HSV-1 virions. The final formation of mature virus particles requires the membrane wrapping of tegumented capsids in the cytoplasm, a process termed secondary envelopment. This process involves the coordination of numerous viral and cellular proteins and results in double-membrane structures with enveloped virions contained within cellular vesicles. Mature viruses are then released through the fusion of these virion-containing vesicles and plasma membranes. This thesis describes investigation into the functions of viral glycoprotein M (gM) and the cellular Endosomal Sorting Complexes Required for Transport (ESCRT) in secondary envelopment. Firstly, it has been reported that gH/L can be efficiently internalised and targeted to the TGN by the co-expression of gM in transfection assays. In order to examine the role of gM in guiding the localisation of viral proteins in infected cells, a HSV-1 gM deletion virus (∆gM), and its revertant virus were constructed. The major phenotype demonstrated was that the absence of gM caused the internalisation of cell surface gH/L to be inhibited and higher levels of gH/L to be observed on the cell surface. Further, lower levels of gH/L were detected in purified ∆gM virions, which was in agreement with the delayed entry kinetics, smaller plaque sizes and greater replication deficits at low multiplicity of infection observed in ∆gM infected cells. Over all the results presented in this thesis demonstrate that in infected cells the efficient incorporation of gH/L into virions relies on the function of gM in HSV-1. Secondly, during HSV-1 secondary envelopment the budding and scission of the viral envelope from the host membrane share topological similarities with the formation of intraluminal vesicle in multivesicular bodies, retrovirus budding, and abscission at the end of cytokinesis, processes that require the cellular ESCRT machinery. There are four multiprotein ESCRT complexes and many associated proteins involved in their regulation. It has been previously shown that the ESCRT-III complex and a functional ATPase VPS4 are required for HSV-1 secondary envelopment, but different from the strategy utilised by HIV-1, the recruitment of ESCRT during HSV-1 infection is independent of TSG101 and/or ALIX. Data presented in this thesis demonstrate that CHMP4A/B/C proteins of the ESCRT-III complex are specifically crucial for HSV-1 secondary envelopment. Simultaneous depletion of CHMP4A/B/C proteins significantly inhibited HSV-1 replication. Ultrastructure analysis revealed that there were virtually no extracellular virions in CHMP4A/B/C depleted samples while more free capsids were observed in the cytoplasm, although the nuclear capsids and primary envelopment events appeared to be normal. In order to identify interactions between HSV-1 and ESCRT proteins, 22 HSV-1 tegument proteins were cloned and tested against a panel of ESCRT and ESCRT-associated proteins in yeast two-hydrid assays. Analysis of positive hits from yeast two-hybrid interaction screens using GST pull-down, co-immunoprecipitation and protein co-localisation assays have validated interactions of pUL47 with CC2D1A/1B, CIN85, CHMP6 and ALIX, pUL46 and pUL49 with CC2D1A/1B and CIN85, and pUL16 with CC2D1A/1B. Furthermore, the newly identified ESCRT associated proteins CC2D1A and CC2D1B have been detected in purified virions. The role of the identified ESCRT proteins in HSV-1 replication has been investigated using siRNA depletion. Unfortunately siRNA depletions of the various ESCRT candidates individually or in combinations did not show any significant effect on HSV-1 replication. Overall these data suggest that unlike HIV and other retroviruses, HSV-1 has evolved multiple parallel pathways to hijack the ESCRT machinery to facilitate its replication, particularly, through the interactions that lead directly to the recruitment of CHMP4A/B/C proteins. Disruption of some of these pathways did not prevent HSV-1 replication in tissue culture, suggesting any one potential pathway is sufficient for ESCRT recruitment to sites of HSV-1 assembly.

Age stratified seroepidemiological studies play a crucial role in the design and assessment of vaccination strategies. An existing multiplex bead immunoassay for measles, mumps, rubella and varicella zoster virus antibodies together with a newly developed multiplex bead immunoassay for herpes simplex virus type 1 and type 2 antibodies were used to investigate the age-related seroepidemiology of these viruses in England during 2012.To develop the HSV-1 and HSV-2 antibody assay, attempts were made to produce full length of HSV-1 and HSV-2 glycoprotein G using a baculovirus vector expression system. While HSV-1 gG protein was produced, the proteins were extensively aggregated. Native glycoprotein G molecules undergo partial removal of HSV-1 signal sequence and HSV-1 short membrane anchor sequence during post translational modification. It is possible that such post translational modification is not performed when protein is processed in insect cell culture. Attempts to produce an HSV-2 glycoprotein G were not successful. It is possible that the high GC-content of HSV-2 glycoprotein G led to poor fidelity of copying the PCR amplification sequence. Commercially available truncated HSV-1 gG and HSV-2 gG were therefore used to develop a duplex microbead immunoassay for the simultaneous detection of specific HSV antibodies in human sera. The resultant assays performed with low sensitivity and specificity (HSV-1 of 89% and 66%, respectively and for HSV-2 of 79% and 85%, respectively) compared to the reference HerpeSelect ELISA.The MMRV multiplex bead immunoassay proved rapid, and required minimal sample volume to semi-quantify MMRV specific antibodies. The seroepidemiology of MMR results was compared with previous seroepidemiological studies performed in 1996 in England. The comparison showed an increase in the proportion of individuals who were positive for mumps and measles antibodies in the 2012 survey. The proportion of individuals positive for rubella was essentially unchanged. The increase in the proportion of individuals positive for mumps and measles antibodies in 2012 show the effectiveness of the change in MMR vaccination policy for England from 1996 onward. For VZV, the proportion of individuals who were positive for varicella antibodies between the 1996 and 2012 serological surveys were essentially unchanged. The comparison showed that most young children are susceptible to VZV. At this level of immunity, it can be expected that varicella will continue to produce epidemics of infection in the population, unless varicella vaccination is implemented as a part of routine childhood vaccination.

Cette thèse a un double objectif : 1) Étudier le comportement sémantique et syntaxique des verbes simples persans par rapport aux prédicats complexes dans les événements spatiaux respectant la typologie de Talmy sur cadrage verbal ou cadrage satellitaire ; et 2) Analyser la systématicité sémantique dans la formation des prédicats complexes basée sur une approche constructionnelle et expliquer leurs comportements exacts, à savoir leur productivité, la causalité, leur durée, et le comportement syntaxique de leurs pré-verbes en fonction de leurs significations. Dans la première partie de l’analyse, nous parlons du comportement des verbes simples par rapport aux prédicats complexes dans les événements spatiaux. Notre analyse est basée sur un vaste corpus. Les prédicats complexes expriment fréquemment le comportement et d’autres informations sémantiques tandis qu’une fréquence élevée de verbes simples apparaît comme des constructions neutres sans exprimer aucune information spécifique. La distribution syntaxique des prédicats complexes respecte la règle de l’« économie de la langue » étant donné que toutes les informations sémantiques sont aussi encodées dans la construction elles-mêmes. Par ailleurs, les verbes simples exigent, dans la majorité des cas, des clauses dépendantes pour exprimer le comportement. En outre, les prédicats complexes encodent, à la fois, le mouvement et les événements d’emplacement, tandis que les verbes simples sont pour la plupart intransitifs et se réfèrent aux événements locatifs. Les lacunes mentionnées ci-dessus sur des verbes simples ont conduit au remplacement de ces derniers par des prédicats complexes au fil du temps. Dans la deuxième partie de notre analyse, nous traitons les caractéristiques sémantiques de prédicats complexes dans des événements spatiaux à la lumière de la théorie de Grammaire de Construction et des approches fondées sur le corpus. Selon Goldberg, nous considérons des constructions de prédicats complexes à part entière. Kardan, "Faire" ; dādan, 'Donner'; zadan, 'Frapper' ; et bordan, 'Prendre', sont les verbes supports étudiés dans cette thèse. Ces verbes supports sont parmi ceux les plus fréquents ayant le plus de pré-verbes en commun dans notre corpus. Premièrement, nous avons identifié les extensions sémantiques de chaque verbe support. Cette catégorisation nous a permis de révéler un certain degré de productivité sémantique ainsi que la généralisation entre chaque verbes supports et le type de pré-verbes avec lesquels il pourrait co-se produire. Autrement dit, chaque verbe support se combine avec un type spécifique de pré-verbe. Il est à note que même des extensions métaphoriques ont un certain degré de productivité et les locuteurs peuvent ainsi les généraliser. Nous tenons également à noter que les prédicats complexes dans les événements spatiaux font partis de la famille de constructions résultatives. Certains comportements des prédicats complexes, tels que leur causalité et la durée de l’activité, sont expliqués en fonction de leur sémantique, et il est démontré que toute la construction est responsable de tels comportements. La comparaison entre les combinaisons qui ont des verbes supports différents mais des pré-verbes identiques évoque quelques différences sémantiques et syntaxiques, ce qui est la preuve de la systématicité existante dans la formation des prédicats complexes. Grâce à une analyse collostructionnelle, nous parlons du degré d’attraction de chaque verbe support par rapport aux catégories sémantiques. Enfin, le type de fréquences de prédicats complexes révèle le degré de productivité dans chaque verbe support, qui ressemble à: Kardan=Zadan > dādan > bordan. / The purpose of this dissertation is twofold : 1) to investigate the semantic and syntactic behavior of Persian simplex verbs vs. CPs in spatial events with respect to Talmy’s typology of satellite-framed and verb-framed languages ; and 2) to analyze the semantic systematicity in the formation of CPs based on a constructional approach and explain their certain behaviors, namely, their productivity, causativity, duration, and syntactic behavior of their PVs through the meaning of the construction.In the first part of the analysis, we discuss the behaviour of simplex verbs vs. CPs in spatial events. Our analysis is based on an extensive corpus. CPs express manner and other semantic information frequently while a high frequency of simplex verbs appears as neutral constructions without expressing any specific information. The syntactic distribution of CPs follow the ‘economy of language’ rule since all the semantic information is encoded in the construction itself; on the other hand, simplex verbs require dependent clauses to express manner in the majority of cases. Furthermore, CPs encode both motion and location events while simplex verbs are mostly intransitive and refer to locative events. The above-mentioned shortcomings of simplex verbs have led to the replacement of these verbs by CPs over time. In the second part of our analysis, we deal with the semantic features of CPs in spatial events in the light of Construction Grammar theory and corpus-based approaches. Following Goldberg we consider CPs constructions in their own right. Kardan, ‘do’ ; dādan, ‘give’; zadan, ‘hit’; and bordan, ‘take’, are the LVs under the study in this thesis. These LVs found to be among the most frequent ones with the most PVs in common in our corpus. Firstly, we identified the semantic extensions of each LV. This categorization reveals a certain degree of semantic productivity and generalization between each LV and the type of PVs with which it can co-occur. In other words, each LV combines with a specific type of PVs. Even metaphorical extensions have a certain degree of productivity and speakers can generalize about them. We also argue that CPs in spatial events are a family of resultative constructions. Certain behaviors of CPs, namely, their causativity, and the duration of the activity, are explained based on their semantics, and it is shown that the whole construction is responsible for such behaviors. The comparison between combinations that have different LVs but identical PVs reveals certain semantic and syntactic differences, which is evidence for the existing systematicity in the formation of CPs. Through a collostructional analysis we discuss the degree of attraction of each LV with respect to the semantic categories.

After initially being infected with a virus, before an adaptive immune response can be mounted, the innate immune system of a cell recognizes and responds to certain patterns present in pathogenic molecules. I studied the role of two genes—PIR1 and CD200R1—on the innate immune responses in two different mouse models of viral infection, infection with the picornavirus EMCV (encephalomyocarditis virus) and infection with HSV-1 (herpes simplex virus) in a mouse model of herpes simplex encephalitis, respectively. PIR1 is a putative RNA phosphatase that has been shown to play an important role in antiviral small RNA processing in C. elegans. It has also been shown to interact with the RIG-I-like receptor LGP2 in preliminary mammalian experiments. I sought to characterize the effect PIR1 has on the innate immune response to the virus EMCV in mice. By developing a PIR1-null mouse, I have found that the role of PIR1 in the progression of EMCV in mice is limited. However, in vitro studies show that PIR1 might play an important role in regulating foreign RNA recognition during the earliest time points post-infection. CD200R1 is an anti-inflammatory signaling molecule that is expressed on myeloidderived cells, and whose ligand is highly expressed within the central nervous system. I investigated the role of this receptor in an intracranial model of herpes simplex encephalitis. CD200R1KO mice show improved survival following direct intracranial infection with HSV. I found this increased survival can be attributed to decreased levels of viral replication in CD200R1KO compared to wild-type mice. Further investigation has shown that CD200R1 affects the signaling and upregulation of the pattern-recognition receptor TLR-2 (toll-like receptor 2), and thus CD200R1 may impact HSV-1 replication by affecting TLR2 signaling.

The innate immune system provides the first line of defense against infection. Pathogens are detected though a variety of Pattern Recognition Receptors (PRRs), which activate downstream signaling cascades. Effector molecules such as cytokines and chemokines are released upon activation and aid in cell recruitment, control of pathogen replication, and coordination of the adaptive immune response. Nucleic acids that are released into the cytosol during viral and bacterial infection are recognized through a special class of PRRs, coined cytosolic nucleic acid sensors. Upon recognition, these receptors induce the production of type I interferons and other cytokines to aid in pathogen clearance. Although many cytosolic nucleic acid sensors have been discovered, it is unclear how they work in concert to mediate these responses. The Interferon Gamma Inducible protein (IFI)16 and its proposed mouse orthologue IFI204 are cytosolic DNA sensors that have been linked to the detection of cytosolic DNA during infection with Herpes Simplex Virus (HSV-1). IFI16 binds dsDNA that has been released into the cytosol during viral infection and engages the adaptor molecule Stimulator of Interferon Genes (STING) leading to TANK binding kinase-1 (TBK1) dependent phosphorylation of interferon regulatory factor 3 (IRF3) and transcription of type I interferons and interferon stimulated genes. In addition to its role as a sensor, in chapter two of this thesis we describe a broader role for IFI16 in the regulation of the type I IFN response to RNA and DNA viruses in anti-viral immunity. In an effort to better understand the role of IFI16 in coordinating type I IFN gene regulation, we generated cell lines with stable knockdown of IFI16 and examined responses to DNA and RNA viruses as well as other inducers of IFN such as cyclic-dinucleotides. As expected, stable knockdown of IFI16 led to a severely attenuated type I IFN response to cytosolic DNA ligands and DNA viruses. In contrast, expression of the NF-κB regulated cytokines such as IL-6 and IL-1β were unaffected in IFI16 knockdown cells, suggesting that the role of IFI16 in sensing these triggers was unique to the type I IFN pathway. Surprisingly, we also found that knockdown of IFI16 led to a severe attenuation of expression of IFN-α and IFN stimulated genes such as RIG-I in response to cyclic GMP-AMP (cGAMP), a second messenger produced in response to cGAS, as well as RNA ligands and viruses. Analysis of IFI16 knockdown cells revealed compromised occupancy of RNA polymerase II on the IFN-α promoter in IFI16 knockdown cells suggesting that transcription of ISGs is dependent on IFI16. Since IFI16 knockdown compromised not only DNA virus driven pathways, we propose additional regulatory roles outside of DNA sensing. Collectively, these results indicate that IFI16 plays a role in the regulation of type I IFN gene transcription and production in response to both RNA and DNA viruses. The role of IFI16/IFI204 has been studied extensively in vitro, however the role of the receptors in vivo has yet to be determined. In chapter three of this thesis, we developed a mouse deficient in IFI204 to explore the role of IFI204 in in vivo immune responses to viruses. We investigated the ability of IFI204 deficient cells to induce type I interferons and other cytokines in response to a panel of DNA and RNA ligands in vitro. IFI204 deficient BMDMs displayed a partial defect in type I interferon induction in response to both DNA and RNA ligands and viruses as compared to WT mice. We also observed that this phenotype is time dependent, since there was no change in type I interferon induction after 12 hours post infection as compared to earlier time points. In contrast to these results, expression of the NF-κB regulated cytokines IL-6 and IL-1β were unaffected in IFI16 knockdown cells. These results suggest that IFI204 plays a partial role in the induction of type I interferons in response to both DNA and RNA ligands. Additionally, IFI204 may work in tandem with other receptors in a sequential manner to amplify the type I interferon response. We also studied the involvement of IFI204 in an in vivo model of HSV-1 infection. IFI204 knockout mice produce less brain and serum IFN-β, IL-6, and IL-1β 72 hours post intraperitoneal infection with HSV-1. Furthermore, IFI204 -/- mice are more susceptible to HSV-1 infection as compared to WT mice. These data indicate that IFI204 mediates the response to HSV-1 in vivo by inducing the production of cytokines that are necessary for the control of viral infection.

No description available.

An accurate satellite attitude and orbit control is a key factor for a successful mission. It guarantees for example sun acquisition on solar panels, fine pointing for optimal telescope usage or satellite lifting to reach higher orbits, when required. Furthermore attitude and orbit control is applied to compensate any occurring disturbances within the space environment. The problem tackled in the present thesis is the optimization of thruster commanding to perform spacecraft orbital manoeuvres. The main objective is to develop different algorithms that are suitable for on-board implementation and to compare their performance. For an optimal thruster command distribution the algorithms shall solve linear programming (or optimization) problems, more exact they shall compute thruster on-times to generate desired torques and/or forces, which are requested by the on-board software. In total three different algorithms are developed of which the first one is based on the pseudoinverse of a matrix, the second one is a variation of the Simplex method and the third one is based on Karmarkar’s algorithm, which belongs to the interior-point methods. The last two methods are well known procedures to solve linear programming problems and in theory they have been analyzed before. However this paper proves their practical application and industrial feasibility for orbital manoeuvres of the weather satellites of ESA’s MTG project and their scalability to any number of thrusters on a generic satellite for 6 degrees of freedom manoeuvres. There are 6 MTG satellites and each has 16 one-sided reaction control thrusters, placed at specific positions and pointed towards defined directions. Physical mechanisms limit the thrusters output to minimum on- and off-times. The focus of this thesis will be on the orbital transfer mode, due to the high disturbances that arise during four motor firing sessions at the apogee, executed to reach higher orbits and finally GEO. The firing sessions are performed by a liquid apogee engine and while this engine is in boost mode, the thrusters shall be used for attitude control only. The technique (nominal case) developed by OHB for this maneuver and currently operational uses 4 thrusters only, which are all pointing in the engine’s direction. They are also used to settle the fuel before the engine is turned on. For control the Pseudoinverse method is applied. If one of the 4 thrusters fails, the backup scenario takes place, which includes using 4 totally different thrusters and no fuel settling, due to their unfavorable position with respect to the engine. The initial idea of this work was to develop a controller for 6 thrusters, using only 2 of the 4 nominal case thrusters, to have a better control performance in the backup case. The Pseudoinverse method was developed by OHB before, thus only small changes needed to be applied to work with 6 thrusters. The two other algorithms, based on the Simplex and Karmarkar method, were completely developed from scratch. To analyze their performance several tests were executed. This includes unit tests on a simple computer hardware with different input, Monte Carlo simulations on a cluster to test if the algorithms are suitable for MTG orbital manoeuvres and the application to 12 thrusters, mounted on a generic satellite to generate torques and forces at the same time for 6 degrees of freedom manoeuvres. For each thruster configuration the worst case outputs are shown in so called minimum control authority plots. The performance analysis consists of the maximum and average deviation between requested and generated torque/force, the average computed thruster on-times, the algorithms computation(running) time and iteration steps. For MTG the test results clearly confirm that the usage of 6 thrusters leads to more accurate generated torques and better control authority, than using only 4 thrusters. The Simplex method stands out here in particular, showing excellence performance regarding torque precision. Nevertheless the accuracy goes at the expense of computation effort. While the Pseudoinverse method is very fast and needs only one iteration step, the Simplex is half a magnitude, the Karmarkar one magnitude slower. But the latter lead to lower thruster on-times in terms of firing duration and thus fuel consumption is reduced. Also it is shown that Simplex and Karmarkar can control 12 thrusters at the same time to generate torques and forces, which proves their scalability to any thruster distribution. In the end it comes to the question whether generating a more accurate torque/force or the computational effort, which is strongly hardware dependent, is more important. A decision which depends on the mission’s objective. This paper shows that all three implemented algorithms are able to handle attitude control in the MTG backup scenario and beyond.

Capable and well-organized data mining algorithms are essential and fundamental to helpful, useful, and successful knowledge discovery in databases. We discuss several data mining algorithms including genetic algorithms (GAs). In addition, we propose a modified multivariate Newton's method (NM) approach to data mining of technical data. Several strategies are employed to stabilize Newton's method to pathological function behavior. NM is compared to GAs and to the simplex evolutionary operation algorithm (EVOP). We find that GAs, NM, and EVOP all perform efficiently for well-behaved global optimization functions with NM providing an exponential improvement in convergence rate. For local optimization problems, we find that GAs and EVOP do not provide the desired convergence rate, accuracy, or precision compared to NM for technical data. We find that GAs are favored for their simplicity while NM would be favored for its performance.

The Support Vector Machine (SVM) is a popular binary classification model due to its superior generalization performance, relative ease-of-use, and applicability of kernel methods. SVM training entails solving an associated quadratic programming (QP) that presents significant challenges in terms of speed and memory constraints for very large datasets; therefore, research on numerical optimization techniques tailored to SVM training is vast. Slow training times are especially of concern when one considers that re-training is often necessary at several values of the models regularization parameter, C, as well as associated kernel parameters. The active set method is suitable for solving SVM problem and is in general ideal when the Hessian is dense and the solution is sparse–the case for the `1-loss SVM formulation. There has recently been renewed interest in the active set method as a technique for exploring the entire SVM regularization path, which has been shown to solve the SVM solution at all points along the regularization path (all values of C) in not much more time than it takes, on average, to perform training at a single value of C with traditional methods. Unfortunately, the majority of active set implementations used for SVM training require positive definite kernels, and those implementations that do allow semi-definite kernels tend to be complex and can exhibit instability and, worse, lack of convergence. This severely limits applicability since it precludes the use of the linear kernel, can be an issue when duplicate data points exist, and doesn’t allow use of low-rank kernel approximations to improve tractability for large datasets. The difficulty, in the case of a semi-definite kernel, arises when a particular active set results in a singular KKT matrix (or the equality-constrained problem formed using the active set is semidefinite). Typically this is handled by explicitly detecting the rank of the KKT matrix. Unfortunately, this adds significant complexity to the implementation; and, if care is not taken, numerical iii instability, or worse, failure to converge can result. This research shows that the singular KKT system can be avoided altogether with simple modifications to the active set method. The result is a practical, easy to implement active set method that does not need to explicitly detect the rank of the KKT matrix nor modify factorization or solution methods based upon the rank. Methods are given for both conventional SVM training as well as for computing the regularization path that are simple and numerically stable. First, an efficient revised simplex method is efficiently implemented for SVM training (SVM-RSQP) with semi-definite kernels and shown to out-perform competing active set implementations for SVM training in terms of training time as well as shown to perform on-par with state-of-the-art SVM training algorithms such as SMO and SVMLight. Next, a new regularization path-following algorithm for semi-definite kernels (Simple SVMPath) is shown to be orders of magnitude faster, more accurate, and significantly less complex than competing methods and does not require the use of external solvers. Theoretical analysis reveals new insights into the nature of the path-following algorithms. Finally, a method is given for computing the approximate regularization path and approximate kernel path using the warm-start capability of the proposed revised simplex method (SVM-RSQP) and shown to provide significant, orders of magnitude, speed-ups relative to the traditional grid search where re-training is performed at each parameter value. Surprisingly, it also shown that even when the solution for the entire path is not desired, computing the approximate path can be seen as a speed-up mechanism for obtaining the solution at a single value. New insights are given concerning the limiting behaviors of the regularization and kernel path as well as the use of low-rank kernel approximations.