Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (582 results)

  • Thesis
  • Peer Reviewed
UC San Diego Electronic Theses and Dissertations (2006)
In a number of domains in computer vision, machine learning and psychology, it is common to model and analyze data in terms of pairwise interactions between data elements, that is, distances between pairs of points. A large variety of supervised and unsupervised algorithms exist for learning the structure of such data. However, it is not always the case that interactions between entities can be described in terms of pairs; often, higher order relations of size three and higher offer a more natural formulation. Consider the k-lines clustering problem, where the objective is to group data points in a d- dimensional vector space into k clusters where elements in each cluster are well approximated by a line. As every pair of data points trivially define a line, there does not exist a useful measure of similarity between pairs of points for this problem. However, it is possible to define measures of similarity over triplets of points that indicate how close they are to being collinear. Another example is the task of constructing an embedding obtained from psychometric experiments involving paired comparisons. A common paired comparison experiment is where subjects are shown three images -- X, Y and Z -- and asked to report which of X or Y is more similar to Z. Here each observation is a triadic relation on the set of images. We refer to the problem of learning from such data -sets as higher order learning. In this dissertation we present solutions to two problems in higher order learning. The first is a new technique for clustering called Clique Averaging, in which we use a novel generative model for hypergraphs to construct a clustering algorithm that has better theoretical as well as empirical performance than existing algorithms. We show applications to illumination invariant clustering and k-subspace learning. The second is a new non-metric multi-dimensional scaling (MDS) algorithm for ordinal data, that is, data sets in which instead of the magnitude, only the relative or rank order of distances is known. We will show how this algorithm can be used to construct a perceptual space for reflectance
    Cover page: Learning from higher order relations
    • Thesis
    • Peer Reviewed
    UC San Diego Electronic Theses and Dissertations (2009)
    Managing power consumption and improving energy efficiency is a key driver in the design of computing devices today. This is true for both battery powered mobile devices as well as mains-powered desktop PCs and servers. In case of mobile devices, the focus of optimization is on energy efficiency to maximize battery lifetime. In case of mains- powered devices, we seek to optimize power consumption to reduce energy costs, thermal and environmental concerns. Traditionally, there are two main mechanisms to improve energy efficiency in systems: slowdown techniques that seek to reduce processor speed or radio power against the rate of work done, and shutdown techniques that seek to shut down specific components or subsystems - such as processor, radio, memory - to reduce power used by these components when not in use. The adverse effect of using these techniques is either reduced performance (e.g., increase in latency) and/or usability or loss of functionality. The thesis behind this dissertation is that improved energy efficiency can be achieved through system architectures that seek to design and exploit "collaboration" among heterogeneous but functionally similar subsystems. For instance, multiple radio interfaces with different power/performance characteristics can collaborate to provide an energy- efficient wireless communication subsystem. Furthermore, we show that in systems where such heterogeneity is not naturally present, we can introduce heterogeneous components to improve overall energy efficiency. We show that using collaboration, individual subsystems and even entire platforms can be shut down more aggressively to reduce energy consumption, while reducing adverse impacts on performance or usability. We have used collaboration to do energy efficient operation in several contexts. For battery powered mobile devices we show that wireless radios are the dominant power consumers, and then describe several techniques that use various heterogeneous radios present on these devices in a collaborative manner to improve their battery lifetime substantially, on average by two to three times and in some cases up to 8 times. First we present "Cell2Notify", a technique in which a lower power radio is used purely to wakeup a higher power radio. Next, we present "CoolSpots" and "SwitchR", systems that build a hierarchy of collaborative radios to choose the most appropriate radio interface, taking into account application characteristics as well as various energy and performance metrics. In the case of wall-powered desktop and laptop Personal Computers (PCs) we show that the dominant power consumers are the processors themselves. We then describe "Somniloquy", an architecture that augments a PC with a separate low power secondary processor that can perform some of the functions of the host PC on its behalf. We show that by using the primary processor (i.e. the PC) collaboratively with the secondary processor we can shut down PCs opportunistically, and as a result reduce the overall energy consumption by up to 80% in most cases
      Cover page: Building aggressively duty-cycled platforms to achieve energy efficiency
      • Thesis
      • Peer Reviewed
      UC San Diego Electronic Theses and Dissertations (2012)
      It is well known among the scientific community that mammalian promoters are enriched for histone H3 lysine 4 methylation (H3K4) specifically trimethylation of the lysine (H3K4me3). Such a definition for transcriptional enhancers, regulatory elements in the genome that positively contribute to the transcriptional activity of a particular promoter, has only been discovered recently. Earlier studies from our lab which were further corroborated by studies from other groups show that transcriptional enhancers in mammalian cells show the presence of H3K4me1,another modification of the same lysine (H3K4), but generally lack H3K4me3. It has also been well known that MLL family of lysine methyltransferases (KMTs) which are specific for H3K4 are the enzymes responsible for this H3K4me3. Although, earlier studies have shown that MLL family of KMTs are also capable of methylating H3K4 to H3K4me1 and H3K4me2, their presence at enhancers or their roles in regulation of transcriptional enhancers has not been elucidated completely. Other families of proteins, H3K4 specific demethylases, have recently been discovered to reverse this methylation of H3K4, which for a long time was considered an irreversible reaction. In our quest to identify the proteins that might have an important role in regulation of enhancers, specifically H3K4 monomethylation at enhancers, I started with the identification of the lysine methyltransferases and related proteins that bind to enhancers (Chapter 1). We found that MLLs, JARID1C and LSD1 are recruited to transcriptional enhancers in human cells. In Chapter 2, we do knock-down experiments on LSD1 and JARID1C to see the effect of their loss on gene expression. We find LSD1 acts as a repressor whose loss results in upregulation of a significant number of genes whereas JARID1C's loss results in upregulation and downregulation of similar number of genes. We also conduct GWLA of LSD1 in knock-down cells to show LSD1's loss result in a genome-wide increase in H3K27 acetylation and H3K4me3 at both promoters and enhancers. In Chapter 3, we study the role of LSD1 in DNA methylation and control of transcription from repetitive DNA elements in the mouse genome. Chapter 4 represents an unrelated study where we studied binding patterns of several transcription factors and histone modifications to identify cis-regulatory modules in the human genome
        Cover page: Regulation of mammalian transcriptional enhancers by H3K4 methyltransferases and H3K4 demethylases
        • Article
        • Peer Reviewed
        UC San Francisco Previously Published Works (2021)
        Anemia is common in patients with chronic kidney disease (CKD) and results from the dysregulation of iron metabolism and erythropoiesis. Hepcidin is a key regulator of iron availability and leads to iron sequestration during the state of iron repletion. Decreases in the level of hepcidin in the presence of hypoxia and/or iron limitation allow for greater iron availability for erythropoiesis. However, kidney excretion of hepcidin decreases as the severity of CKD increases, whereas production of hepcidin is increased under inflammatory conditions often present in patients with CKD, both of which contribute to anemia. Assessment of iron status is, therefore, essential in the treatment of anemia. However, current laboratory tests for the determination of the adequate supply of iron have many limitations, including diurnal variation in the levels of biomarkers, lack of standardized reference methods across laboratories, and confounding by the presence of inflammation. In addition, the current treatment paradigm for anemia of CKD can further disrupt iron homeostasis; for example, treatment with erythropoiesis-stimulating agents in the absence of supplemental iron can induce functional iron deficiency. Moreover, supplemental iron can further increase levels of hepcidin. Several novel therapies, including hypoxia-inducible factor prolyl hydroxylase inhibitors and hepcidin inhibitors/antagonists, have shown promise in attenuating the levels and/or activity of hepcidin in anemia of CKD, thus ensuring the availability of iron for erythropoiesis.
          Cover page: Iron metabolism and management: focus on chronic kidney disease.
          • Thesis
          • Peer Reviewed
          UC San Diego Electronic Theses and Dissertations (2015)
          In this dissertation we discuss some conjectured dualities in supersymmetric field theories and provide non-trivial checks for these conjectures. A quick review of supersymmetry and related topics is provided in chapter 1. In chapter 2, we develop a method to identify the so called BPS states in the Hilbert space of a supersymmetric field theory (that preserves at least two real supercharges) on a generic curved space. As an application we obtain the superconformal index (SCI) of 4d theories. The large N SCI of quiver gauge theories has been previously noticed to factorize over the set of extremal BPS mesonic operators. In chapter 3, we reformulate this factorization in terms of the zigzag paths in the dimer model associated to the quiver and extend the factorization theorem of the index to include theories obtained from D-branes probing orbifold singularities. In chapter 4, we consider the dualities in two classes of 3 dimensional theories. The first class consist of dualities of certain necklace type Chern-Simons (CS) quiver gauge theories. A non trivial check of these dualities is provided by matching their squashed sphere partition functions. The second class consists of theories whose duals are described by a collection of free fields. In such cases, due to mixing between the superconformal R- symmetry and accidental symmetries, the matching of electric and magnetic partition functions is not straightforward. We provide a prescription to rectify this mismatch. In chapter 5, we consider some the N = 1 4d theories with orthogonal and symplectic gauge groups, arising from N = 1 preserving reduction of 6d theories on a Riemann surface. This construction allows us to dual descriptions of 4d theories. Some of the dual frames have no known Lagrangian description. We check the dualities by computing the anomaly coefficients and the superconformal indices. We also give a prescription to write the index of the theory obtained by reduction of 6d theories on a three punctured sphere with Z₂ and Z₃ twist lines and verify that it exhibits the conjectured symmetry enhancement from G₂ USp(6) to E7. In chapter 6, we continue our study of 4d theories obtained from reduction of 6d theories. We introduce a new type of object that we call the 'Fan' and show how to construct new N = 1 superconformal theories using the Fan. In chapter 7, we demonstrate the existence of an infinite number of theories that are either dual to or exhibit a cascade of Rug flows down to the Sou(N) Sq cd with four flavors and a quartic superpotential
            Cover page: Dualities and Curved Space Partition Functions of Supersymmetric Theories
            • Article
            LBL Publications (1999)

            International Monitoring System (IMS) stations and the International Data Centre (IDC) of the Preparatory Commission for the Comprehensive Nuclear-Test-Ban Treaty Organization generate data and products that must be transmitted to one or more receivers. The application protocols used to transmit the IMS data and IDC products will be CD-x and IMS-x and the World Wide Web (WWW). These protocols use existing Internet applications and Internet protocols to send their data. The primary Internet applications in use are electronic mail (e-mail) and the file transfer protocol (ftp). The primary Internet communication protocol in use is the Transmission Control Protocol (TCP), which provides reliable delivery to the receiver. These Internet applications and protocol provide unicast (point-to-point) communication. A message sent using unicast has a single recipient; any message intended for more than one recipient must be sent to each recipient individually. In the current design, the IDC and the National Data Centres (NDC's) provide data forwarding to the appropriate receivers. The overhead associated with using unicast to transmit messages to multiple receivers either directly or through a forwarder increases linearly with the number of receivers. In addition, using a forwarding site introduces possible delays and possible points of failure in the path to the receivers. Reliable multicast provides communication services similar to TCP but for a group of receivers. The reliable multicast protocol provides group membership services and message delivery ordering. If an IMS station were to send its data using reliable multicast instead of unicast, only sites that are members of the multicast group would receive the data at approximately the same time. This might provide an efficient means of disseminating station data or IDC data products to all receivers and eliminate or greatly reduce the need for data forwarding. Several commercial and research reliable multicast protocols exist for the Internet. Each of these protocols is designed to serve a specific community of users and applications. The author has undertaken a study to determine if reliable multicasting is appropriate for use in the Global Communication Infrastructure (GCI).

              Cover page: Using multicast in the global communications infrastructure for group communication
              • Article
              LBL Publications (2009)

              The Fluxnet synthesis dataset originally compiled for the La Thuile workshop contained approximately 600 site years. Since the workshop, several additional site years have been added and the dataset now contains over 920 site years from over 240 sites. A data refresh update is expected to increase those numbers in the next few months. The ancillary data describing the sites continues to evolve as well. There are on the order of 120 site contacts and 60 proposals have been approved to use the data. These proposals involve around 120 researchers. The size and complexity of the dataset and collaboration has led to a new approach to providing access to the data and collaboration support and the support team attended the workshop and worked closely with the attendees and the Fluxnet project office to define the requirements for the support infrastructure. As a result of this effort, a new website (http://www.fluxdata.org) has been created to provide access to the Fluxnet synthesis dataset. This new web site is based on a scientific data server which enables browsing of the data on-line, data download, and version tracking. We leverage database and data analysis tools such as OLAP data cubes and web reports to enable browser and Excel pivot table access to the data.

                Cover page: Fluxnet Synthesis Dataset Collaboration Infrastructure
                • Article
                LBL Publications (1976)
                  Cover page: MEASUREMENT OF N02 CONCENTRATION IN COMBUSTION USING FLUORESCENCE EXCITED BY AN ARGON-ION LASER
                  • Article
                  • Peer Reviewed
                  Lawrence Berkeley National Laboratory (2010)

                  Carbon-climate, like other environmental sciences, has been changing. Large-scale synthesis studies are becoming more common. These synthesis studies are often conducted by science teams that are geographically distributed and on datasets that are global in scale. A broad array of collaboration and data analytics tools are now available that could support these science teams. However, building tools that scientists actually use is hard. Also, moving scientists from an informal collaboration structure to one mediated by technology often exposes inconsistencies in the understanding of the rules of engagement between collaborators. We have developed a scientific collaboration portal, called fluxdata.org, which serves the community of scientists providing and analyzing the global FLUXNET carbon-flux synthesis dataset. Key things we learned or re-learned during our portal development include: minimize the barrier to entry, provide features on a just-in-time basis, development of requirements is an on-going process, provide incentives to change leaders and leverage the opportunity they represent, automate as much as possible, and you can only learn how to make it better if people depend on it enough to give you feedback. In addition, we also learned that splitting the portal roles between scientists and computer scientists improved user adoption and trust. The fluxdata.org portal has now been in operation for ~;;1.5 years and has become central to the FLUXNET synthesis efforts.

                    Cover page: A DATA-CENTERED COLLABORATION PORTAL TO SUPPORT GLOBAL CARBON-FLUX ANALYSIS
                    • Article
                    • Peer Reviewed
                    California Breast Cancer Research Program Funded Publications (2014)
                      Cover page: FGF receptor genes and breast cancer susceptibility: results from the Breast Cancer Association Consortium
                      Top