Virginia Tech - Wake Forest University School of Biomedical Engineering & Sciences
Computer Science Technical Reports @Virginia TechNot a member yet
997 research outputs found
Sort by
Cross-Platform Presentation of Interactive Volumetric Imagery
Full text linkVolume data is useful across many disciplines, not just medicine.
Thus, it is very important that researchers have a simple and
lightweight method of sharing and reproducing such volumetric
data. In this paper, we explore some of the challenges associated
with volume rendering, both from a classical sense and from the
context of Web3D technologies. We describe and evaluate the pro-
posed X3D Volume Rendering Component and its associated styles
for their suitability in the visualization of several types of image
data. Additionally, we examine the ability for a minimal X3D node
set to capture provenance and semantic information from outside
ontologies in metadata and integrate it with the scene graph
An SMP Soft Classification Algorithm for Remote Sensing
Full text linkThis work introduces a symmetric multiprocessing (SMP) version of the continuous iterative
guided spectral class rejection (CIGSCR) algorithm, a semiautomated classiï¬cation algorithm for remote
sensing (multispectral) images. The algorithm uses soft data clusters to produce a soft classiï¬cation
containing inherently more information than a comparable hard classiï¬cation at an increased computational
cost. Previous work suggests that similar algorithms achieve good parallel scalability, motivating the parallel
algorithm development work here. Experimental results of applying parallel CIGSCR to an image with
approximately 10^8 pixels and six bands demonstrate superlinear speedup. A soft two class classiï¬cation is
generated in just over four minutes using 32 processors
A fully discrete framework for the adaptive solution of inverse problems
Full text linkWe investigate and contrast the differences between the discretize-then-differentiate and differentiate-then-discretize approaches to the numerical solution of parameter estimation problems. The former approach is attractive in practice due to the use of automatic differentiation for the generation of the dual and optimality equations in the first-order KKT system. The latter strategy is more versatile, in that it allows one to formulate efficient mesh-independent algorithms over suitably chosen function spaces. However, it is significantly more difficult to implement, since automatic code generation is no longer an option. The starting point is a classical elliptic inverse problem. An a priori error analysis for the discrete optimality equation shows consistency and stability are not inherited automatically from the primal discretization. Similar to the concept of dual consistency, We introduce the concept of optimality consistency. However, the convergence properties can be restored through suitable consistent modifications of the target functional. Numerical tests confirm the theoretical convergence order for the optimal solution. We then derive a posteriori error estimates for the infinite dimensional optimal solution error, through a suitably chosen error functional. This estimates are constructed using second order derivative information for the target functional. For computational efficiency, the Hessian is replaced by a low order BFGS approximation. The efficiency of the error estimator is confirmed by a numerical experiment with multigrid optimization
Device-Based Isolation for Securing Cryptographic Keys
Full text linkIn this work, we describe an eective device-based isolation
approach for achieving data security. Device-based isolation
leverages the proliferation of personal computing devices to
provide strong run-time guarantees for the condentiality of
secrets. To demonstrate our isolation approach, we show its
use in protecting the secrecy of highly sensitive data that
is crucial to security operations, such as cryptographic keys
used for decrypting ciphertext or signing digital signatures.
Private key is usually encrypted when not used, however,
when being used, the plaintext key is loaded into the memory
of the host for access. In our threat model, the host may
be compromised by attackers, and thus the condentiality of
the host memory cannot be preserved. We present a novel
and practical solution and its prototype called DataGuard to
protect the secrecy of the highly sensitive data through the
storage isolation and secure tunneling enabled by a mobile
handheld device. DataGuard can be deployed for the key
protection of individuals or organizations
A class of implicit-explicit two-step Runge-Kutta methods
Full text linkThis work develops implicit-explicit time integrators based on two-step Runge-Kutta methods.
The class of schemes of interest is characterized by linear invariant
preservation and high stage orders. Theoretical consistency and stability analyses are performed to reveal the properties of these methods. The new framework offers extreme flexibility
in the construction of partitioned integrators, since no coupling conditions are necessary.
Moreover, the methods are not plagued by severe order reduction, due to their high stage orders.
Two practical schemes of orders four and six are constructed, and are used to solve several test problems.
Numerical results confirm the theoretical findings
User Intention-Based Traffic Dependence Analysis For Anomaly Detection
Full text linkThis paper describes an approach for enforcing
dependencies between network traffic and user activities for
anomaly detection. We present a framework and algorithms that
analyze user actions and network events on a host according
to their dependencies. Discovering these relations is useful in
identifying anomalous events on a host that are caused by
software flaws or malicious code. To demonstrate the feasibility
of user intention-based traffic dependence analysis, we
implement a prototype called CR-Miner and perform extensive
experimental evaluation of the accuracy, security, and efficiency
of our algorithm. The results show that our algorithm can
identify user intention-based traffic dependence with high accuracy
(average 99:6% for 20 users) and low false alarms. Our
prototype can successfully detect several pieces of HTTP-based
real-world spyware. Our dependence analysis is fast with a
minimal storage requirement. We give a thorough analysis on
the security and robustness of the user intention-based traffic
dependence approach
CoreTSAR: Task Scheduling for Accelerator-aware Runtimes
Full text linkHeterogeneous supercomputers that incorporate computational accelerators
such as GPUs are increasingly popular due to their high
peak performance, energy efficiency and comparatively low cost.
Unfortunately, the programming models and frameworks designed
to extract performance from all computational units still lack the
flexibility of their CPU-only counterparts. Accelerated OpenMP
improves this situation by supporting natural migration of OpenMP
code from CPUs to a GPU. However, these implementations currently
lose one of OpenMP’s best features, its flexibility: typical
OpenMP applications can run on any number of CPUs. GPU implementations
do not transparently employ multiple GPUs on a node
or a mix of GPUs and CPUs. To address these shortcomings, we
present CoreTSAR, our runtime library for dynamically scheduling
tasks across heterogeneous resources, and propose straightforward
extensions that incorporate this functionality into Accelerated
OpenMP. We show that our approach can provide nearly linear
speedup to four GPUs over only using CPUs or one GPU while
increasing the overall flexibility of Accelerated OpenMP
Towards Energy-Proportional Computing for Enterprise-Class Server Workloads
Full text linkMassive data centers housing thousands of computing nodes
have become commonplace in enterprise computing, and the
power consumption of such data centers is growing at an
unprecedented rate. Adding to the problem is the inability
of the servers to exhibit energy proportionality, i.e., provide
energy-ecient execution under all levels of utilization,
which diminishes the overall energy eciency of the data
center. It is imperative that we realize eective strategies
to control the power consumption of the server and improve
the energy eciency of data centers. With the advent of
Intel Sandy Bridge processors, we have the ability to specify
a limit on power consumption during runtime, which creates
opportunities to design new power-management techniques
for enterprise workloads and make the systems that they run
on more energy-proportional.
In this paper, we investigate whether it is possible to achieve
energy proportionality for an enterprise-class server workload,
namely SPECpower ssj2008 benchmark, by using Intel's
Running Average Power Limit (RAPL) interfaces. First,
we analyze the power consumption and characterize the instantaneous
power prole of the SPECpower benchmark at
a subsystem-level using the on-chip energy meters exposed
via the RAPL interfaces. We then analyze the impact of
RAPL power limiting on the performance, per-transaction
response time, power consumption, and energy eciency of
the benchmark under dierent load levels. Our observations
and results shed light on the ecacy of the RAPL interfaces
and provide guidance for designing power-management techniques
for enterprise-class workloads
Nonreciprocating Sharing Methods in Cooperative Q-Learning Environments
Full text linkPast research on multiagent simulation with cooperative reinforcement learning (RL) focuses on developing sharing strategies that are adopted and used by all agents in the environment. In this paper, we target situations where this assumption of a single sharing strategy that is employed by all agents is not valid. We seek to address how agents with no predetermined sharing partners can exploit groups of cooperatively learning agents to improve learning performance when compared to independent learning. Specifically, we propose three intra-agent methods that do not assume a reciprocating sharing relationship and leverage the pre-existing agent interface associated with Q-Learning to expedite learning
Performance Analysis of a Novel GPU Computation-to-core Mapping Scheme for Robust Facet Image Modeling
Full text linkThough the GPGPU concept is well-known
in image processing, much more work remains to be done
to fully exploit GPUs as an alternative computation
engine. This paper investigates the computation-to-core
mapping strategies to probe the efficiency and scalability
of the robust facet image modeling algorithm on GPUs.
Our fine-grained computation-to-core mapping scheme
shows a significant performance gain over the standard
pixel-wise mapping scheme. With in-depth performance
comparisons across the two different mapping schemes,
we analyze the impact of the level of parallelism on
the GPU computation and suggest two principles for
optimizing future image processing applications on the
GPU platform