WICKRAMARATHNE, T. L. (M.S., Electrical and Computer Engineering) A Belief Theoretic Approach for Automated Collaborative Filtering (May 2008) Abstract of a thesis at the University of Miami. Thesis supervised by Professor Kamal Premaratne. No. of pages in text. (84) Automated Collaborative Filtering (ACF) is one of the most successful strategies available for recommender systems. Application of ACF in more sensitive and critical applications however has been hampered by the absence of better mechanisms to accommodate imperfections (ambiguities and uncertainties in ratings, missing ratings, etc.) that are inherent in user preference ratings and propagate such imperfections throughout the decision making process. Thus one is compelled to make various "assumptions" regarding the user preferences giving rise to predictions that lack sufficient integrity. With its Dempster-Shafer belief theoretic basis, CoFiDS, the automated Collaborative Filtering algorithm proposed in this thesis, can (a) represent a wide variety of data imperfections; (b) propagate the partial knowledge that such data imperfections generate throughout the decision-making process; and (c) conveniently incorporate contextual information from multiple sources. The "soft" predictions that CoFiDS generates provide substantial exibility to the domain expert. Depending on the associated DS theoretic belief-plausibility measures, the domain expert can either render a "hard" decision or narrow down the possible set of predictions to as smaller set as necessary. With its capability to accommodate data imperfections, CoFiDS widens the applicability of ACF, from the more popular domains, such as movie and book recommendations, to more sensitive and critical problem domains, such as medical expert support systems, homeland security and surveillance, etc. We use a benchmark movie dataset and a synthetic dataset to validate CoFiDS and compare it to several existing ACF systems.

An array of ship types was used during the European Age of Expansion (early 15th to early 17th centuries), but one vessel in particular emerges from the historical records as a harbinger of discovery: the caravel. The problem is that little is known about these popular ships of discovery, despite the fair amount of historical evidence that has been uncovered. How big were they? How many men did it take to operate such a vessel? What kind of sailing characteristics did they have? How and by whom were they designed? Where did they originate and how did they develop? These questions cannot be answered by looking at the historical accounts alone. For this reason, scholars must take another approach for learning about caravels by examining additional sources, namely ancient shipbuilding treatises, archaeological evidence, surviving archaic shipbuilding techniques, and iconographic representations from the past. Information gained from the available sources reveals many of the caravel’s characteristics through time. This ship type outclassed its contemporaries during the age of exploration because of its highly adaptive characteristics. These traits were, principally, its shallow draught, speed, maneuverability, and ability to sail close to the wind. This combination of attributes made the caravel the ideal ship for reconnaissance along the rocky African coastline, as well as for making the transatlantic voyages to the New World. It was built in a Mediterranean way during its post-medieval phases, a method that still survives in some parts of the world today. During the Age of Discovery (ca. 1430 to 1530), the caravel sat low in the water, had one sterncastle, and was either lateen-rigged or had a combination of square and lateen sails. This vessel reflects the advanced shipbuilding technology that existed in Europe at this time, and played and important role in the voyages which allowed the Europeans to expand their territories around the world. The results of the studies presented in this thesis provide a history and development of the caravel, which was gradual and often obscure. What has been gained from this work is a body of information that can be applied to other studies about ancient seafaring, and can serve as a starting point for further research.

The process of drug discovery from natural products starts with the selection of study object. In this project recent knowledge and methods are incorporated to investigate the process of such selection for pharmacognostic investigations. As the model and object of study mistletoes and their content of the small cytotoxic peptides thionins are chosen. The thionins are compared in silico to other proposed plant innate defense peptides. Utilizing analysis of amino acid sequences and secondary structures, the thionins are shown to be one of eight distinct groups of cystein-rich plant polypeptides analysed. Common features of thionins are exploited in an investigation of isolation methods, where a simple acidic extraction is equally efficient to isolate thionins as the laborious methods hitherto used. An effort to study the relationships of the order Santalales was done. To infer phylogenetic relationships from DNA sequences, we increased the taxon sampling for utilized genes and regions such as rbcL, atpB and ribosomal 18S and 26S rDNA sequences within the Santalales. Analysing these together with published sequences for other tricolpate taxa a position for Santalales as sister to caryophyllids and basal to asterids is implied. This indication is supported by chemical characters such as the presence of cyclopeptide alkaloids of a kind only known from Gentianales. To validate the chemosystematic implications from thionin distribution extracts of mistletoes collected in Panama, Taiwan and Madagascar, and the relative Osyris alba (Santalaceae) collected in Spain, were screened with the established fluorescence microculture cytotoxicity assay using the thionin-sensitive human lymphoma cell-line U937GTB. Bioassay guided isolation concludes that the cytotoxic compounds in Loranthaceae may however constitute another group of peptides. In conclusion this work shows that the incorporation of informatic techniques may aid prediction and decision making when planning pharmacognostic research.

No description available.

The Bluetooth Scatternet Formation (BSF) problem can be defined as the problem of forming wireless networks of Bluetooth devices in an efficient manner. A number of restrictions imposed by the Bluetooth specifications make the BSF problem challenging and unique. Many interesting solution algorithms have been proposed in the literature to solve this problem. In this thesis, we investigate the BSF problem. We concentrate on problems introduced by the procedures of device discovery of the Bluetooth specifications and on the different solutions used by BSF algorithms to deal with these problems. We study also in this thesis problems introduced by the specifications of link establishment in Bluetooth due to their close interaction with the device discovery specifications. We survey and categorize the different device discovery techniques used by BSF algorithms. This categorization is then used as a basis to identify the different theoretical computational models used to study BSF algorithms. We argue, in this thesis, that the currently available models for Bluetooth wireless networks do not model adequately, in most cases, the complexities of the Bluetooth specifications and we show that these models were oversimplified in many cases. A general computational model will be useful as a starting point to design BSF algorithms and to compare the different and numerous BSF algorithms – especially in term of the execution time efficiency. In this thesis, we provide a set of suggestions that will help in the creation of such model. We survey a number of studies that examined in more depth the specifications of device discovery in Bluetooth. We survey also other studies that attempted to simplify the Bluetooth network model, either by suggesting modifications on the Bluetooth specifications or by the use of communication technologies other than Bluetooth. Finally, we present some experiments accompanied with analyzes to show the complexities of the Bluetooth specifications and their sensitivity to minor changes (whether in the specifications or in their implementation).

The success of the internet over last few decades has mainly depended on various infrastructure technologies to run distributed applications. Due to diversification and multi-provider nature of the internet, radical architectural improvements which require mutual agreement between infrastructure providers have become highly impractical. This escalating resistance towards the further growth has created a rising demand for new approaches to address this challenge. Network virtualization is regarded as a prominent solution to surmount these limitations. It decouples the conventional Internet service provider’s role into infrastructure provider (InP) and service provider (SP) and introduce a third player known as virtual network Provider (VNP) which creates virtual networks (VNs). Resource discovery aims to assist the VNP in selecting the best InP that has the best matching resources for a particular VN request. In the current literature, resource discovery focuses mainly on static attributes of network resources highlighting the fact that utilization on dynamic attributes imposes significant overhead on the network itself. In this thesis we propose a resource discovery approach that is capable of utilizing the dynamic resource attributes to enhance the resource discovery and increase the overall efficiency of VN creation. We realize that recourse discovery techniques should be fast and cost efficient, enough to not to impose any significant load. Hence our proposed scheme calculates aggregation values of the dynamic attributes of the substrate resources. By comparing aggregation values to VN requirements, a set of potential InPs is selected. The potential InPs satisfy basic VN embedding requirements. Moreover, we propose further enhancements to the dynamic attribute monitoring process using a vector based aggregation approach.

In this body of work, we present the details of a novel method for passive resource discovery in cluster grid environments where resources constantly utilize inter-node communication. Our method offers the ability to non-intrusively identify resources that have available memory or CPU cycles; this is critical for lowering queue wait times in large cluster grid networks, and for memory-intensive cluster grid applica-tions such as Gaussian (computational chemistry package) and the Weather Research and Forecasting (WRF) modeling package. The benefits include: (1) low message complexity, (2) scalability, (3) load bal-ancing support, and (4) low maintainability. Using several test-beds (i.e., a small local test-bed and a 50-node Deterlab test-bed), we demonstrate the feasibility of our method with experiments utilizing TCP, UDP and ICMP network traffic. Using this technique, we observed a correlation between memory or CPU load and the timely response of network traffic. In such situations, we have observed that in highly utilized (due to multi-programming) nodes there will be numerous, active processes which require context switching or paging. The latency associated with numerous context switches or paging manifests as a de-lay signature within the packet transmission process. Our method detects this delay signature to determine the utilization of network resources. The aforementioned delay signature is the keystone that provides a correlation between network traffic and the internal state of the source node. We characterize this delay signature due to CPU utilization by (1) identifying the different types of assembly language instructions that source this delay and (2) describing how performance-enhancing techniques (e.g., instruction pipelin-ing, caching) impact this delay signature by using the LEON3, implemented as a 40 MHz development board. At the software level, results for medium sized networks show that our method can consistently and accurately identify nodes with available memory or CPU cycles (< 70% availability). At the hardware level, our results show that excessive context switching in active applications increases the average mem-ory access time, thus adding additional delay to the execution of LD instructions. Additionally, internal use of these instructions in heavily utilized situations to send network packets induces the delay signature into network traffic.

Drug development research on wound repair is challenging and inefficient due to the complex nature of wound healing and scarring processes and the limitations of available in vitro or in vivo models used for preclinical drug testing. Many patients who undergo elective back surgery develop post-surgical complications resulting from excess peridural scarring in and around the site of operation. We tested the effects of two anti-inflammatory compounds, quercetin and L-2-oxothiazolidine-4-carboxylate (OTC), in ameliorating peridural scar formation following spinal laminectomy surgery in laboratory rats. Western blot and immunocytochemical analyses indicated that the peridural scar tissue contained MyoD-positive myoblast cells and expressed prolyl-4-hydroxylase (P4H), a fibroblast marker. Treatment with 1 mM OTC reduced activation of ERK1/2 and p38 mitogen-activated protein kinases (MAPK) at 21 days post-surgery suggesting potential anti-scarring mechanism. However, large animal to animal variation in the expression levels of collagen biosynthesis markers made it difficult to demonstrate any efficacy of quercetin or OTC in reducing peridural scar formation. The shortcomings of this live animal approach led us to develop a novel three-dimensional (3-D) <i>in vitro</i> wound repair model for evaluating quercetin and OTC effects. High-density micromass co-cultures seeded at a 1:3 ratio of FR 3T3 fibroblast cells and L8 myoblast cells formed 3-D microtissues <i>in vitro</i> that expressed MyoD, P4H, and á-smooth muscle actin. The micromass tissue layer remained adherent to the culture plate when inflicted with a single laceration injury, which allowed monitoring of cell migration into the wound site. Wounded cultures were treated with quercetin, OTC and other agents (TGF- â1, mitomycin, p38 inhibitor SB202190, ERK inhibitor PD184352) to determine their effects on collagen accumulation, wound closure rates, MAPK activation, and gene transcript expression. Both OTC and quercetin treatments reduced collagen biosynthesis in dose-dependent manner. In addition, 1.5 mM OTC accelerated wound closure and significantly reduced p38 MAPK activation without affecting ERK1/2. In contrast, 40 µM quercetin delayed wound closure in micromass co-cultures and reduced ERK1/2 activation. Our in vitro findings suggest that OTC might have potential as an anti-scarring agent. Importantly, our novel micromass co-culture system shows promise as an improved 3-D scaffold-free in vitro model for use in preclinical drug development research.

A majority of kinase inhibitors predominantly occupy the highly conserved adenine-binding pocket located in the kinase catalytic cleft, and therefore the target selectivity of these molecules is a major concern. In order to design highly specific next-generation drugs, it is essential to exploit the less-conserved binding pockets, which lie adjacent to the adenine-binding pocket. Small peptides that can function as adenosine triphosphate (ATP) competitive inhibitors would prove useful in identifying and validating new druggable surfaces in the kinase catalytic cleft. These peptides, being larger than small molecules, have the potential to target the ATP binding pocket as well as surfaces that lie adjacent to this pocket. Such peptides recognizing novel binding pockets can assist the drug discovery process in several ways. In this thesis, we describe the isolation and characterization of a novel class of cyclic peptides, referred to as lariats, against Abl kinase, a drug target important in chronic myeloid leukemia and other disorders. Using a yeast two-hybrid approach, we first isolated two related lariats, named A1 and A2, from a pool of five million lariats, which interact with the catalytic domain of Abl kinase. In vitro studies indicated that the synthetic A1 lariat competitively inhibits ATP binding by targeting the catalytic cleft that lies between the N- and C- lobes of the kinase catalytic domain. To obtain tighter-binding variants of the A1 lariat, we developed an affinity maturation protocol consisting of two steps. In the first step, we defined acceptable and tolerable substitutions at each position of the A1 lariat using site-saturation mutagenesis (SSM). In the second step, we designed specific mutations to the A1 lariat based on the SSM results and evolved higher affinity variants. Synthetic and recombinant higher affinity lariats exhibited a strong inhibition of Abl kinase activity in vitro and Bcr-Abl kinase activity in vivo, respectively, illustrating the potential of lariats as chemical genetic tools. Resistance mutation profiling showed that the lariats are not affected by the activating mutations located in the activation loop of kinase, and instead bind preferentially to the kinase active conformation. Selectivity analysis indicated that the lariats do not recognize Src family kinases, which share a high structural similarity with Abl kinase in their active conformation. These findings, coupled with preliminary results from modeling studies, strongly suggest that the lariats have identified novel allosteric drug-binding pockets in the kinase catalytic cleft.

The genome-wide HaploInsufficieny Profiling (HIPHOP) technique has been validated as a method to quantify the relative abundance of uniquely tagged yeast deletion strains using a microarray readout. The massive throughput of next generation sequencing presents a new technology for assessing HIPHOP profiles. I developed a new method called Barcode analysis by Sequencing (Bar-seq) that applies deep sequencing to genome-scale fitness. I show that Bar-seq outperforms the current benchmark barcode microarray assay in terms of both dynamic range and throughput. When applied to a complex genome-scale fitness assay, Bar-seq quantitatively identifies drug-targets, exceeding the performance of the microarray assay. I also established that Bar-seq is well suited to a multiplex format and provides a dramatic increase in throughput. I used the genome-wide HIPHOP assay and other functional genomics tools to explore the mechanisms underlying drug-drug synergies. Drug combination therapy, and synergistic combinations in particular, have several advantages over monotherapies. Synergistic drug combinations allow the dose of each agent to be reduced, often with the benefit of diminishing side effects while maintaining efficacy and decreasing the chances of drug resistance. I used my yeast model to identify synergistic drug combinations and found that inhibitors of ergosterol biosynthesis are highly synergistic with several agents, including those targeting other points within the same pathway. I also devised a method that enriches for synergistic interactions during screening of compound combinations. This new synergy prediction method can aid in the rapid identification of anti-proliferative combinations and can be readily applied to other organisms for further characterization and/or confirmation. Finally, I examined synergistic combination HIPHOP profiles and identified Gene Ontology enrichments that are combination-specific.