I have a PHD. from langara community college. I spent 4 years picking up trash in Mongolia to prevent starving homeless children in the middle east to find their true passion. I currently reside in Florida, South Canada with my 19 children and 4 wives. I got mad bitches fr.
Industrial & Engineering Chemistry Research, Oct 3, 2007
ABSTRACT We present results of our experimental study on the effect of particle surface coverage ... more ABSTRACT We present results of our experimental study on the effect of particle surface coverage (density) on the evolution of packing structure in charged granular assemblies in a two-dimensional (2D) geometry. Steel particles were charged by rubbing them with the bottom surface of a polystyrene box. We conducted a Monte Carlo simulation study via the interparticle interaction of the dipole repulsion. The resulting radial distribution functions, bond orientation correlation functions, and 2D pressure isotherm approximate the experimental data well. The particle potential of the mean force changes from repulsion only at a low particle coverage fraction of 0.01 to oscillation between repulsion and attraction at higher coverage fractions due to the collective particle interactions.
Proceedings on Privacy Enhancing Technologies, Jul 1, 2023
A large body of research has shown that machine learning models are vulnerable to membership infe... more A large body of research has shown that machine learning models are vulnerable to membership inference (MI) attacks that violate the privacy of the participants in the training data. Most MI research focuses on the case of a single standalone model, while production machine-learning platforms often update models over time, on data that often shifts in distribution, giving the attacker more information. This paper proposes new attacks that take advantage of one or more model updates to improve MI. A key part of our approach is to leverage rich information from standalone MI attacks mounted separately against the original and updated models, and to combine this information in specific ways to improve attack effectiveness. We propose a set of combination functions and tuning methods for each, and present both analytical and quantitative justification for various options. Our results on four public datasets show that our attacks are effective at using update information to give the adversary a significant advantage over attacks on standalone models, but also compared to a prior MI attack that takes advantage of model updates in a related machine-unlearning setting. We perform the first measurements of the impact of distribution shift on MI attacks with model updates, and show that a more drastic distribution shift results in significantly higher MI risk than a gradual shift. We also show that our attacks are effective at auditing differentially private fine tuning. We make our code public on Github: https://github.com/stanleykywu/model-updates.
In vivo studies of mice were performed to investigate whether auto-reactive antibodies specific f... more In vivo studies of mice were performed to investigate whether auto-reactive antibodies specific for self CD20 antigen on B cells could be induced by immunizing with a CD20 peptide linked to a foreign, human IgG.Fc fragment through a T cell immunologically inert linker peptide and how such an auto-reactivity, if generated, would affect the levels of B cells. The dimeric Fc fusion protein containing the extracellular 44-amino acid portion of CD20, and the CH2-CH3 domains of human gamma 1 immunoglobulin were prepared. After several subcutaneous immunizations with this CD20-Fc protein, mice produced anti-CD20 antibodies that can bind to native CD20 on normal B cells and B-lymphoma cells. In mice immunized with the CD20-Fc protein, the fraction of B cells in total peripheral blood lymphocytes decreased to about 40%, significantly lower than that of mice immunized with human IgG. In addition, antibody response towards an irrelevant bystander antigen, chicken ovalbumin, was weakened compared with that of mice immunized with human IgG. These results show that auto-reactive antibodies specific for CD20 can be induced by immunizing with an autologous CD20 peptide fused with a foreign IgG.Fc and that the auto-antibodies can partially reduce the levels of B cells and their response to other antigens.
Abstract Myelin basic protein (MBP) has a well-known association with multiple sclerosis. However... more Abstract Myelin basic protein (MBP) has a well-known association with multiple sclerosis. However, it is not well-studied as a biomarker for traumatic brain injury (TBI). To date there are only small studies which look at the level of MBP in traumatic brain injury but the data is promising and compelling enough to push for further larger studies. Calcium has been implicated as having an association with myelin damage, but again there is not enough compelling evidence to give support for making any conclusions. Another promising area of research with MBP and TBI is in neuroimaging and measuring myelin water as a surrogate for myelin. Here again, larger studies will need to be performed to make any positive conclusions.
Use of experimentally derived induced pluripotent stem cells (iPSCs) has led to the development o... more Use of experimentally derived induced pluripotent stem cells (iPSCs) has led to the development of cell models for differentiation, drug testing and understanding disease pathogenesis. For these models to be informative, reprogrammed cell lines need to be adequately characterized and shown to preserve all of the critical characteristics of pluripotency and differentiation. Here, we report a detailed protocol for the generation of iPSCs from human fibroblasts containing mutations in COL3A1 using a Sendai virus mediated integration-free reprogramming approach. We describe how to characterize the putative iPSCs in vivo and in vitro to ensure potency and differentiation potential. As an example of how these mutations may affect cell surface and extracellular matrix (ECM) interactions, we provide protocols for the differentiation of these cells into smooth muscle cells to illustrate how different cell types may display cell autonomous differences in collagen receptors that may affect their phenotype. These cells, when applied to mechanical model systems (see Chapter 18 by Bose et al.) facilitate an assessment of stiffness and stress-strain relationships useful for understanding how extracellular matrix dysfunction and its interactions with surface proteins contribute to disease processes.
Abstract We studied the self-organization of different-sized particles in different confined geom... more Abstract We studied the self-organization of different-sized particles in different confined geometries. The structure formations of millimeter-sized granules confined to a solid substrate, micrometer-sized colloidal particles confined to a fluid-liquid interface, and ...
Industrial & Engineering Chemistry Research, Oct 3, 2007
ABSTRACT We present results of our experimental study on the effect of particle surface coverage ... more ABSTRACT We present results of our experimental study on the effect of particle surface coverage (density) on the evolution of packing structure in charged granular assemblies in a two-dimensional (2D) geometry. Steel particles were charged by rubbing them with the bottom surface of a polystyrene box. We conducted a Monte Carlo simulation study via the interparticle interaction of the dipole repulsion. The resulting radial distribution functions, bond orientation correlation functions, and 2D pressure isotherm approximate the experimental data well. The particle potential of the mean force changes from repulsion only at a low particle coverage fraction of 0.01 to oscillation between repulsion and attraction at higher coverage fractions due to the collective particle interactions.
Proceedings on Privacy Enhancing Technologies, Jul 1, 2023
A large body of research has shown that machine learning models are vulnerable to membership infe... more A large body of research has shown that machine learning models are vulnerable to membership inference (MI) attacks that violate the privacy of the participants in the training data. Most MI research focuses on the case of a single standalone model, while production machine-learning platforms often update models over time, on data that often shifts in distribution, giving the attacker more information. This paper proposes new attacks that take advantage of one or more model updates to improve MI. A key part of our approach is to leverage rich information from standalone MI attacks mounted separately against the original and updated models, and to combine this information in specific ways to improve attack effectiveness. We propose a set of combination functions and tuning methods for each, and present both analytical and quantitative justification for various options. Our results on four public datasets show that our attacks are effective at using update information to give the adversary a significant advantage over attacks on standalone models, but also compared to a prior MI attack that takes advantage of model updates in a related machine-unlearning setting. We perform the first measurements of the impact of distribution shift on MI attacks with model updates, and show that a more drastic distribution shift results in significantly higher MI risk than a gradual shift. We also show that our attacks are effective at auditing differentially private fine tuning. We make our code public on Github: https://github.com/stanleykywu/model-updates.
In vivo studies of mice were performed to investigate whether auto-reactive antibodies specific f... more In vivo studies of mice were performed to investigate whether auto-reactive antibodies specific for self CD20 antigen on B cells could be induced by immunizing with a CD20 peptide linked to a foreign, human IgG.Fc fragment through a T cell immunologically inert linker peptide and how such an auto-reactivity, if generated, would affect the levels of B cells. The dimeric Fc fusion protein containing the extracellular 44-amino acid portion of CD20, and the CH2-CH3 domains of human gamma 1 immunoglobulin were prepared. After several subcutaneous immunizations with this CD20-Fc protein, mice produced anti-CD20 antibodies that can bind to native CD20 on normal B cells and B-lymphoma cells. In mice immunized with the CD20-Fc protein, the fraction of B cells in total peripheral blood lymphocytes decreased to about 40%, significantly lower than that of mice immunized with human IgG. In addition, antibody response towards an irrelevant bystander antigen, chicken ovalbumin, was weakened compared with that of mice immunized with human IgG. These results show that auto-reactive antibodies specific for CD20 can be induced by immunizing with an autologous CD20 peptide fused with a foreign IgG.Fc and that the auto-antibodies can partially reduce the levels of B cells and their response to other antigens.
Abstract Myelin basic protein (MBP) has a well-known association with multiple sclerosis. However... more Abstract Myelin basic protein (MBP) has a well-known association with multiple sclerosis. However, it is not well-studied as a biomarker for traumatic brain injury (TBI). To date there are only small studies which look at the level of MBP in traumatic brain injury but the data is promising and compelling enough to push for further larger studies. Calcium has been implicated as having an association with myelin damage, but again there is not enough compelling evidence to give support for making any conclusions. Another promising area of research with MBP and TBI is in neuroimaging and measuring myelin water as a surrogate for myelin. Here again, larger studies will need to be performed to make any positive conclusions.
Use of experimentally derived induced pluripotent stem cells (iPSCs) has led to the development o... more Use of experimentally derived induced pluripotent stem cells (iPSCs) has led to the development of cell models for differentiation, drug testing and understanding disease pathogenesis. For these models to be informative, reprogrammed cell lines need to be adequately characterized and shown to preserve all of the critical characteristics of pluripotency and differentiation. Here, we report a detailed protocol for the generation of iPSCs from human fibroblasts containing mutations in COL3A1 using a Sendai virus mediated integration-free reprogramming approach. We describe how to characterize the putative iPSCs in vivo and in vitro to ensure potency and differentiation potential. As an example of how these mutations may affect cell surface and extracellular matrix (ECM) interactions, we provide protocols for the differentiation of these cells into smooth muscle cells to illustrate how different cell types may display cell autonomous differences in collagen receptors that may affect their phenotype. These cells, when applied to mechanical model systems (see Chapter 18 by Bose et al.) facilitate an assessment of stiffness and stress-strain relationships useful for understanding how extracellular matrix dysfunction and its interactions with surface proteins contribute to disease processes.
Abstract We studied the self-organization of different-sized particles in different confined geom... more Abstract We studied the self-organization of different-sized particles in different confined geometries. The structure formations of millimeter-sized granules confined to a solid substrate, micrometer-sized colloidal particles confined to a fluid-liquid interface, and ...
Uploads