Search Results (1 - 16 of 16 Results)

Sort By  
Sort Dir
 
Results per page  

Howard, Shaun MichaelDeep Learning for Sensor Fusion
Master of Sciences (Engineering), Case Western Reserve University, 2017, EECS - Computer and Information Sciences
The use of multiple sensors in modern day vehicular applications is necessary to provide a complete outlook of surroundings for advanced driver assistance systems (ADAS) and automated driving. The fusion of these sensors provides increased certainty in the recognition, localization and prediction of surroundings. A deep learning-based sensor fusion system is proposed to fuse two independent, multi-modal sensor sources. This system is shown to successfully learn the complex capabilities of an existing state-of-the-art sensor fusion system and generalize well to new sensor fusion datasets. It has high precision and recall with minimal confusion after training on several million examples of labeled multi-modal sensor data. It is robust, has a sustainable training time, and has real-time response capabilities on a deep learning PC with a single NVIDIA GeForce GTX 980Ti graphical processing unit (GPU).

Committee:

Wyatt Newman, Dr (Committee Chair); M. Cenk Cavusoglu, Dr (Committee Member); Michael Lewicki, Dr (Committee Member)

Subjects:

Artificial Intelligence; Computer Science

Keywords:

deep learning; sensor fusion; deep neural networks; advanced driver assistance systems; automated driving; multi-stream neural networks; feedforward; multilayer perceptron; recurrent; gated recurrent unit; long-short term memory; camera; radar;

Shao, YuanlongLearning Sparse Recurrent Neural Networks in Language Modeling
Master of Science, The Ohio State University, 2014, Computer Science and Engineering
In the context of statistical language modeling, we explored the task of learning an Elman network with sparse weight matrices, as a pilot study towards learning a sparsely con-nected fully recurrent neural network, which would be potentially useful in many cases. We also explored how efficient and scalable it can be in practice. In particular, we explored these tasks: (1) We adapted the Iterative Hard Thresholding (IHT) algorithm into the BackPropagation Through Time (BPTT) learning. (2) To accel-erate convergence of the IHT algorithm, we designed a scheme for expanding the net-work by replicating the existing hidden neurons. Thus we can start training from a small and dense network which is already learned. (3) We implemented this algorithm in GPU. Under small minibatch sizes and large network sizes (e.g., 2000 hidden neurons) it achieves 160 times speedup compared to the RNNLM toolkit in CPU. With larger mini-batch sizes there could be another 10 times speedup, though the convergence rate be-comes an issue in such cases and further effort is needed to address this problem. (4) Without theoretical convergence guarantee of the IHT algorithm in our problem setting, we did an empirical study showing that learning a sparse network does give competitive perplexity in language modeling. In particular, we showed that a sparse network learned in this way can outperform a dense network when the number of effective parameters is kept the same. (5) We gathered performance metric comparing the computational effi-ciency of the matrix operations of interest in both sparse and dense settings. The results suggest that for network sizes which we can train in reasonable time at this moment, it’s hard for sparse matrices to run faster, unless we are allowed to have very sparse networks. Thus for research purposes we may want to focus on using dense matrices, while for en-gineering purposes a more flexible matrix design leveraging the power of dense and sparse matrices might be necessary.

Committee:

Eric Fosler-Lussier, Dr. (Advisor); Mikhail Belkin, Dr. (Committee Member)

Subjects:

Artificial Intelligence; Computer Science

Keywords:

language modeling; recurrent neural networks; sparse recurrent neural networks

Putchala, Manoj KumarDeep Learning Approach for Intrusion Detection System (IDS) in the Internet of Things (IoT) Network using Gated Recurrent Neural Networks (GRU)
Master of Science (MS), Wright State University, 2017, Computer Science
The Internet of Things (IoT) is a complex paradigm where billions of devices are connected to a network. These connected devices form an intelligent system of systems that share the data without human-to-computer or human-to-human interaction. These systems extract meaningful data that can transform human lives, businesses, and the world in significant ways. However, the reality of IoT is prone to countless cyber-attacks in the extremely hostile environment like the internet. The recent hack of 2014 Jeep Cherokee, iStan pacemaker, and a German steel plant are a few notable security breaches. To secure an IoT system, the traditional high-end security solutions are not suitable, as IoT devices are of low storage capacity and less processing power. Moreover, the IoT devices are connected for longer time periods without human intervention. This raises a need to develop smart security solutions which are light-weight, distributed and have a high longevity of service. Rather than per-device security for numerous IoT devices, it is more feasible to implement security solutions for network data. The artificial intelligence theories like Machine Learning and Deep Learning have already proven their significance when dealing with heterogeneous data of various sizes. To substantiate this, in this research, we have applied concepts of Deep Learning and Transmission Control Protocol/Internet Protocol (TCP/IP) to build a light-weight distributed security solution with high durability for IoT network security. First, we have examined the ways of improving IoT architecture and proposed a light-weight and multi-layered design for an IoT network. Second, we have analyzed the existingapplications of Machine Learning and Deep Learning to the IoT and Cyber-Security. Third, we have evaluated deep learning’s Gated Recurrent Neural Networks (LSTM and GRU) on the DARPA/KDD Cup '99 intrusion detection data set for each layer in the designed architecture. Finally, from the evaluated metrics, we have proposed the best neural network design suitable for the IoT Intrusion Detection System. With an accuracy of 98.91% and False Alarm Rate of 0.76 %, this unique research outperformed the performance results of existing methods over the KDD Cup ’99 dataset. For this first time in the IoT research, the concepts of Gated Recurrent Neural Networks are applied for the IoT security.

Committee:

Michelle Cheatham, Ph.D. (Advisor); Adam Bryant, Ph.D. (Committee Member); Mateen Rizki, Ph.D. (Committee Member)

Subjects:

Computer Science

Keywords:

Deep Learning; Internet of Things; Machine Learning; Gated Recurrent Unit; Recurrent Neural Networks

Chen, ChengStudy of Indicators of Recurrent Congestion on Urban Roadway Network Based on Bus Probes
Master of Science, The Ohio State University, 2010, Civil Engineering

Congestion has long been a concern to transportation operation and planning. Various methods and technologies have been developed and applied in detecting congestion. This study focuses on using bus as probes to indicate recurrent congestion on urban roadway networks.

In this study, bus speeds are used to reflect indications of congestion. Bus speed data collected over “homogeneous days” are pooled in order to find indications of recurrent congestion. Bus speeds collected during a time period at each location are compared to bus speeds collected during other time periods (or during the whole day) at the same location. For practical purposes, the entire bus route investigated for the purpose of method development, validation, and demonstration is discretized into 10 meter spatial sections, and time of day is discretized into 30 minute time intervals. Each 10 meter long spatial section and 30 minute duration time interval constitutes a space-time cell for which existence or absence of an indication of congestion is determined.

Three different methods are used to determine the existence or absence of an indication of congestion for each space-time cell. Two of the three methods, the Mean Based method and the Median Based method, apply traditional statistical tests. The other method, the Low Speed Threshold (LST) method, is developed in this study. An empirical study based on Automatic Vehicle Location (AVL) data collected using The Ohio State University (OSU) Campus Transit Lab (CTL) is conducted to evaluate the three methods. The indications of congestion produced by the three methods are then compared to a priori expectation of recurrent congestions along the route.

The Mean Based method produces few indications of recurrent congestion. It misses space-time cells where recurrent congestion is expected. The Median Based method and the LST method produce similar general patterns in terms of where and when recurrent congestions are indicated. However, there exist some differences between the Median Based method and the LST method in terms of the specific locations where and specific times when recurrent congestions are indicated.

A systematic analysis and comparison based on the Median Based method and the LST method is conducted. Based on the empirical analyses and a priori expectation of recurrent congestions along the route, the LST method seems more appealing than the Median Based method (and the Mean Based method) for determining indications of recurrent congestion. Although the empirical results produced by the LST method are encouraging, more research is necessary before concluding that the LST method could be used to detect recurrent congestion on a widespread basis. A sensitivity analysis of the parameters used in this study is suggested for future research. Other traffic related elements such as speed limits, bus trip running times, and schedule adherence might also be utilized to improve the recurrent congestion detection methods.

Committee:

Mark McCord (Advisor); Rabi Mishalani (Advisor); Prem Goel (Committee Member)

Subjects:

Transportation

Keywords:

recurrent congestion; bus probes; urban street

Zale, Peter J.GERMPLASM COLLECTION, CHARACTERIZATION, AND ENHANCEMENT OF EASTERN PHLOX SPECIES
Doctor of Philosophy, The Ohio State University, 2014, Horticulture and Crop Science

The genus Phlox is a staple of gardens worldwide that includes species admired for beauty and versatility in gardens, constructed landscapes, containers, and as cut flowers. Extensive breeding and selection has occurred in three primary species: Phlox drummondii, P. paniculata, and P. subulata, but the genus includes other species with ornamental value. Phlox L. (Polemoniaceae) includes approximately 65 species primarily endemic to North America; 20-23 species occur in the eastern U.S. and 40-45 in the west. The eastern species are a polymorphic group organized into 6 subsections that include the three main cultivated species and up to 20 related, rarely cultivated species. Thus, the species diversity of Phlox has barely been applied for ornamental use. The widespread availability of diverse germplasm can contribute not only to new cultivated forms but also to a greater understanding of species diversity and relationships. Such interest has made Phlox a priority genus for conservation at the Ornamental Plant Germplasm Center. This work describes the development, partial characterization, and manipulation of Phlox germplasm.

Phlox germplasm collection development began in 2010 with an effort to collect all eastern species from natural populations throughout their native ranges; 187 accessions were collected from wild populations of 22 eastern species during a series of expeditions. Another 166 accessions were of cultivated origin; these were used for comparison to wild-collected material. The 353 accessions represent the most comprehensive germplasm collection of Phlox to date.

This germplasm was first characterized by estimation of genome size using flow cytometry and ploidy estimates by chromosome counts. Genome size was surveyed in 287 accessions; 165 accessions were of wild origin and the rest were cultivars. Most accessions were diploid, but genome size was variable; both tetraploid and hexaploid populations were found in a species where it had not been previously reported. The 122 cultivated selections were also primarily diploid, but anueploids were discovered suggesting previous polyploid breeding. These data have important implications in the evolution, adaptation, conservation, and commercial breeding of Phlox.

Molecular characterization included a limited application of microsatellite (SSR) marker analysis of 8 populations, 5 diploid and 3 tetraploid, and 61 individuals from the P. pilosa complex. This species complex had high genetic diversity and moderate population structuring, typical of outcrossing species with gametophytic self - incompatibility. The data indicated that P. pilosa is a rich source of genetic diversity for germplasm collections and enhancement.

Three interspecific hybridization experiments were performed to identify gene pools for Phlox germplasm enhancement; in two studies there was at least one recurring parent and the other was a partial diallel. Interspecific hybridization was most successful among closely related species, and success decreased with increasing phylogenetic distance; however, even crosses between closely related species failed to produce seeds, emphasizing that selection of parental taxa is critical. Success at recreating previously described interspecific hybrids was limited, but at least one was resynthesized. Unlike previous failed attempts at interploid hybridization, some interploid crosses were successful and resulted in morphologically intermediate, aneuploid taxa with intermediate genome sizes to the parents.

Committee:

Pablo Jourdan (Advisor); Mark Bennett (Committee Member); David Francis (Committee Member); John Freudenstein (Committee Member)

Subjects:

Horticulture

Keywords:

Germplasm; Accession; Genome Size; Flow Cytometry; Chromosome Number; Diploid; Tetraploid; Aneuploid; Interploid; Phylogeny; Interspecific Hybridization; Plant Breeding; Recurrent Parent; Partial Diallel; Microsatellite Marker;

Benito González, AnaV1-DERIVED RENSHAW CELLS AND IA INHIBITORY INTERNEURONS DIFFERENTIATE EARLY DURING DEVELOPMENT
Doctor of Philosophy (PhD), Wright State University, 2011, Biomedical Sciences PhD
Locomotor development is dependent on the maturation of spinal cord circuits controlling motor output, but little is known about the development of the spinal interneurons that control motoneuron activity. This study focused on the development of Renshaw cells (RCs) and Ia inhibitory interneurons (IaINs), which mediate recurrent and reciprocal inhibition, respectively, two basic inhibitory circuits for motorneuron control. Both interneurons originate from the same progenitor pool (p1) giving rise to ventral spinal embryonic interneurons denominated V1. V1-derived interneurons (V1-INs) establish local inhibitory connections with ipsilateral motoneurons and express the transcription factor engrailed-1. This characteristic permitted the generation of transgenic mice that were used in this study to genetically label V1 interneuron lineages from embryo to adult. Adult V1-derived Renshaw cells and IaINs share some similar properties, both being inhibitory and establishing ipsilateral connections; but differ in morphology, location in relation to motor pools, expression of calcium binding proteins (calbindin vs. parvabumin), synaptic connectivity and function. These differences are already present in neonates, therefore the purpose of this study was to determine possible embryonic differentiation mechanisms. Using 5‟-bromodeoxyuridine birth-dating we demonstrated that V1-INs can be divided into early and late born groups. The early group quickly upregulates calbindin iv expression and includes the Renshaw cells, which maintain calbindin expression through life. The second group includes many cells that postnatally upregulate parvalbumin, including IaINs. This later born group is characterized by upregulation of the transcription factor FoxP2 as they start to differentiate and is retained up to the first postnatal week in many V1-derived IaINs. In contrast, Renshaw cells express the transcription factor MafB that seems relatively specific to them within the V1-INs. Furthermore, Renshaw cells appear attracted to the ventral root exit region and follow a unique migratory route to become specifically placed at this location. In contrast, other V1 interneurons settle more medially and far from the ventral root exit region. MafB expression is upregulated in Renshaw cells only after they have reached their final position among motor axons. Therefore, the specific migration of Renshaw cells might be responsible for their final differentiation and unique relationship with motor axons in adult.

Committee:

Francisco Alvarez, PhD (Advisor); Paula Bubulya, PhD (Committee Member); Timothy Cope, PhD (Committee Member); David Ladle, PhD (Committee Member); James Olson, PhD (Committee Member)

Subjects:

Neurology

Keywords:

spinal cord; development; neurogenesis; locomotor circuits; V1-interneurons; engrailed-1; calbindin; parvalbumin; ventral horn; recurrent inhibition; reciprocal inhibition; motoneuron

Chen, ChengUsing Archived Bus Automatic Vehicle Location Data to Identify Indications of Recurrent Congestion
Doctor of Philosophy, The Ohio State University, 2013, Civil Engineering
This study focuses on using buses equipped with Automated Vehicle Location (AVL) systems as probes to detect recurrent congestion indications on urban streets. A previously developed method for finding indications of recurrent congestion for a given period of multiple days is used as the base method. An approach is proposed to determine appropriate values for parameters in this base method. The approach is expected to be general and could therefore be adopted for other applications. The set of appropriate values determined in an empirical application are used for subsequent empirical studies to investigate the reasonableness of methods proposed in this study. The approach is based on using two bus AVL data sets that should produce similar congestion indications. Parameter values that produce a high measure of correlation between the two data sets subject to other considerations are selected. A method is developed to determine periods of multiple days such that there is no change in recurrent traffic patterns during a period and that there is a change between different periods. The homogeneous days grouping method consists of two components: bottom-up clustering and re-grouping. The bottom-up clustering component iteratively combines consecutive groups of days that are similar in terms of the speed distributions obtained on days in the groups. Based on the groups of days determined by the bottom-up clustering component, the re-grouping component further combines non-consecutive groups if they are similar in terms of their speed distributions. The method is validated using three different approaches. The three validation approaches all support the promise of the proposed method. An investigation is conducted to determine possible influence of bus drivers’ reactions to schedules on indications of recurrent congestion. Two procedures are developed to handle two different cases of bus drivers’ reactions to schedules: driving at low speed and driving at high speed to adhere to a schedule. Each approach uses multiple criteria to determine time-of-day periods and locations of indications of the corresponding bus drivers’ reactions to schedules. The speeds observed within the time-of-day periods and locations where bus drivers’ reactions are indicated to exist are removed. Empirical results support the usefulness of fine-tuning the recurrent congestion detection method by removing false indications resulting from bus drivers’ reactions to schedules.

Committee:

Rabi Mishalani, Ph.D. (Advisor); Mark McCord, Ph.D. (Advisor); Cathy Xia, Ph.D. (Committee Member)

Subjects:

Civil Engineering; Transportation

Keywords:

recurrent congestion; bus probes; Automatic Vehicle Location; AVL; network sensing; traffic condition assessment

Schuler, Tammy A.Marital Quality Affects Biobehavioral Outcomes in Advanced and Recurrent Breast Cancer Patients
Doctor of Philosophy, The Ohio State University, 2011, Psychology
Advanced and recurrent breast cancer patients experience negative biobehavioral sequelae following diagnosis. Poor marital quality has also been shown to worsen biobehavioral trajectories in earlier-stage cancer patients (e.g., Yang & Schuler, 2009; Schuler et al., under review). However, the contribution of poor marital quality among advanced or recurrent cancer patients coping with a health crisis remains unclear. This study tested the longitudinal covariation between poor marital quality and psychological distress, individual differences, health behaviors, endocrine and immune functioning, and physical health in advanced and recurrent breast cancer patients (N=98). Mixed-effects modeling compared trajectories for women in distressed marriages (n=23) to those in non-distressed marriages (n=75) at diagnosis and across a 12-month follow-up. Compared with patients in a non-distressed marriage, those in a distressed marriage showed significantly greater baseline total mood disturbance (p<.001) and differential rate of mood disturbance change across follow-up (p=.018). Immune differences were also present, with the Distressed group showing significantly higher Con A at baseline relative to the Non-Distressed group (p=.052), which persisted across 12-month follow-up. Clinical relevance and recommendations are described.

Committee:

Barbara Andersen, PhD (Advisor); Daniel Strunk, PhD (Committee Member); Robert Cudeck, PhD (Committee Member); Steven Beck, PhD (Committee Member); Helen Everts, PhD (Other)

Subjects:

Behavioral Psychology; Behavioral Sciences; Behaviorial Sciences; Clinical Psychology; Families and Family Life; Psychobiology; Psychology; Psychotherapy; Social Research

Keywords:

cancer; disease outcomes; psychological distress; biobehavioral; marital; marital distress; marital quality; marital conflict; breast cancer; advanced cancer; recurrent cancer; marital therapy

Bastian, ChinthasagarThe Role of Synaptically Released Free Zinc in the Zinc Rich Region of Epileptic Mammalian Hippocampal Circuitry
Doctor of Philosophy (PhD), Ohio University, 2010, Biological Sciences (Arts and Sciences)

Mammalian hippocampal circuitry contains copious vesicular Zn2+ and glutamate in the mossy fiber axon terminals which originate from the granule cells of the dentate gyrus. Temporal lobe epilepsy in humans and rodent epilepsy models reveal a peculiar feature in the brain, branches referred to as “recurrent mossy fibers” which sprouts off from mossy fibers. The recurrent mossy fiber terminals of pilocarpine induced epileptic rats were examined for Zn2+ and confirmed with Timm's staining and intracellular fluorescent zinc indicators. These zinc-rich terminals were investigated for release of Zn2+ into the extracellular space detected by a low affinity fluorescent indicator for Zn2+, namely Newport Green. This study provides evidence to support that high frequency electrical stimulation of the mossy fiber axons causes Zn2+ release from not only mossy fibers but also the Zn2+ rich recurrent mossy fiber terminals. This release is frequency dependent; the fluorescent response is attenuated in the presence of Zn2+ specific chelators, low calcium medium and a vesicular uptake inhibitor. Also, the concentration of Zn2+ released from the terminals was estimated to be in the low micromolar range, with significantly higher release observed in epileptic animals when compared to sham treated ones. Evidence for a Zn2+ sensitive voltage gated sodium channel, Nav1.5/ SCN5A expression in the dentate gyrus was obtained by western blotting and real time quantitative PCR and using immunostaining they were seen localized to regions around the soma of granule cells and CA3 pyramidal cells in the dentate gyrus.

Interestingly, when epileptic rats were analyzed it was seen that animals that were 16 weeks epileptic had the maximal amount of releasable zinc as well as Nav1.5 protein expression. This was followed by a gradual decline with age. Since these channels could be blocked by Zn2+ and their localization was found near where synaptic Zn2+ is released, it could be postulated that the presence of increased zinc could block Nav1.5 channels and consequently decrease the excitatory activity in the dentate gyrus of epileptic rats. However, a causal effect between the increase in zinc and Nav1.5 channels could not be determined due to the presence of these channels in the age-matched controls. Still, the mere presence of this TTX resistant Zn2+ sensitive channel in the hippocampal pathway makes it an interesting candidate to be studied for the interactions of zinc.

To conclude, this study demonstrates the presence of functional Zn2+ releasing synapses on the anomalous sprouts that are seen in epilepsy, releasing Zn2+ in the low micromolar range. In the epileptic rat hippocampus, the fluorescent response to the released Zn2+ was seen not only in the molecular layer and CA3-hilar region but also observed in the granule cell layer albeit in lower amounts compared to the other areas. This Zn2+ may serve as a contributing factor to several postsynaptic and presynaptic interactions depending on the constituency of the targets in the particular region. Moreover, a potential target for Zn2+, a Zn2+ sensitive channel, SCN5A/Nav1.5 in the hippocampus was identified for the first time.

Committee:

Yang Li, PhD (Advisor); Robert Colvin, PhD (Committee Member); Mark Berryman, PhD (Committee Member); Gary Cordingley, PhD (Committee Member)

Subjects:

Biomedical Research; Molecular Biology

Keywords:

zinc; epilepsy; Nav1.5; SCN5A; hippocampus; recurrent mossy fibers; synaptic release

DeFranco, Emily A.Placental pathologic aberrations in cases of familial idiopathic spontaneous preterm birth
MS, University of Cincinnati, 2010, Medicine: Clinical and Translational Research

OBJECTIVE: To test the hypothesis that placental histologic characteristics in pregnancies complicated by familial spontaneous preterm birth (sPTB) will differ by gestational age (GA) and reflect possible mechanisms of pathogenesis.

METHODS: We conducted a prospective cohort study in women who had both an idiopathic sPTB <35 weeks and a first degree family member affected by PTB. Parturients with clinical chorioamnionitis in labor or medically indicated PTB were excluded. Placental specimens were reviewed by a single pathologist blind to GA at birth. Results were categorized with respect to the presence of maternal and/ or fetal inflammatory response (MIR, FIR). The placental findings were compared to three categories of preterm GAs: 32–35 (referent), 28–32, and <28 weeks, adjusting for statistically influential factors.

RESULTS: Placental specimens were evaluated from 79 spontaneous PTBs. Inflammatory responses were found frequently: 41 (51.9%) had MIR and 28 (35.4%) had FIR. Placental inflammatory changes of maternal origin were most frequent at the earliest GAs, 85% with PTB <28 weeks [adjOR 77.5 (95% CI 5, 1213.1)], and 57% at 32-35 weeks [adjOR 6.1 (0.8, 48.5)] compared to later PTBs occuring at 32–35 weeks (22%). Inflammatory changes of fetal origin (FIR) also occurred more frequently in cases of extreme PTB, adjOR 38.4 (95% CI 2.9, 514.2).

CONCLUSION: Placental inflammatory responses are common in women with familial sPTB. Maternal and fetal inflammatory responses occur most frequently in the earliest cases of PTB. This data suggests that inflammation plays an important role in the onset of parturition in cases otherwise classified as idiopathic or spontaneous in nature, especially at the earliest GAs when neonatal outcomes are the poorest.

Committee:

Erin Nicole Haynes, DrPH (Committee Chair); Louis Muglia, PhD (Committee Member); David Lewis, MD, MBA (Committee Member)

Subjects:

Biostatistics

Keywords:

prematurity;placental pathology;recurrent preterm birth;placental inflammation;preterm labor

Kramer, Gregory RobertAn analysis of neutral drift's effect on the evolution of a CTRNN locomotion controller with noisy fitness evaluation
Doctor of Philosophy (PhD), Wright State University, 2007, Computer Science and Engineering PhD
This dissertation focuses on the evolution of Continuous Time Recurrent Neural Networks (CTRNNs) as controllers for control systems. Existing research suggests that the process of neutral drift can greatly benefit evolution for problems whose fitness landscapes contain large-scale neutral networks. CTRNNs are known to be highly degenerate, providing a possible source of large-scale landscape neutrality, and existing research suggests that neutral drift benefits the evolution of simple CTRNNs. However, there has been no in-depth examination of the effects of neutral drift on complex CTRNN controllers, especially in the presence of noisy fitness evaluation. To address this problem, this dissertation presents an analysis of the effect of neutral drift on the evolution of a complex CTRNN locomotion controller for a simulated hexapod robot in the presence of noisy fitness evaluations. In particular, two stochastic hill-climber-based EAs are examined and compared, one that does not engage in neutral drift, and one that does. The experimental results show that while neutral drift provides a significant advantage early in the evolutionary process, the later effects of noisy fitness evaluations seriously degrades the utility of neutral drift, and overall, there is no significant difference between the non-drifting and drifting EAs. These results provide evidence that large-scale neutral networks do exist in complex CTRNN fitness landscapes and highlight the important role that noisy fitness evaluations play in influencing evolutionary performance.

Committee:

John Gallagher (Advisor)

Subjects:

Computer Science

Keywords:

Continuous Time Recurrent Neural Network; Neutrality; Neutral Drift; Artificial Evolution; Evolutionary Computation; Neural Network; Controller; Evolutionary Algorithm; Noisy Fitness Evaluation

Perumal, SubramoniamStability and Switchability in Recurrent Neural Networks
MS, University of Cincinnati, 2008, Engineering : Computer Science

Artificial Neural Networks (ANNs) are being extensively researched for their wide range of applications. Among the most important is the ability of a type of ANNs—recurrent attractor networks—to work as associative memories. The most common type of ANN used for associative memory is the Hopfield network, which is a fully connected network with symmetric connections. There have been numerous attempts to improve the capacity and recall quality of recurrent networks, with the focus primarily on the stability of the stored attractors, and the network's convergence properties. However, the ability of a recurrent attractor network to switch between attractors is also an interesting property, if it can be harnessed for use. Such switching can be useful as a model of switching between context-dependent functional networks thought to underlie cognitive processing.

In this thesis, we design and develop a stable-yet-switchable (SyS) network model which provides an interesting combination of stability and switchability. The network is stable under random perturbations, but highly sensitive to specific targeted perturbations which cause it to switch attractors. Such functionality has previously been reported in networks with scale-free (SF) connectivity. We introduce networks with two regions: A densely connected core region, and a sparsely connected and larger periphery. We show that these core-periphery (CP) networks are better for providing a combination of stability and targeted switching than scale-free networks. We develop and validate a specific approach to switching between attractors in a targeted way. The CP and SF models are also compared with each other and with randomly connected homogeneous networks.

Committee:

Dr. Ali Minai (Advisor); Dr. Raj Bhatnagar (Committee Member); Dr. Anca Ralescu (Committee Member)

Subjects:

Computer Science; Engineering

Keywords:

recurrent neural networks; core-periphery networks; switchability; switching between attractors; stability and switchability

Murphy, Brian L.Aberrant hippocampal granule cell neurogenesis and integration in epilepsy
PhD, University of Cincinnati, 2010, Medicine: Neuroscience/Medical Science Scholars Interdisciplinary

The data from the present studies suggest that granule cells in different stages of neuronal differentiation may be differentially susceptible to changes in their cellular structure and survival following an epileptogenic stimulus. Recent work in the field has focused on the development of dendritic abnormalities with respect to immature neurons or newly generated neurons following exposure to an epileptogenic stimulus. In Chapter 2, we show for the first time that fully differentiated granule cells are capable of dendritic rearrangement. Specifically, we have shown that previously established apical dendrites shift to the basal portion of the cell as the somata of these cells radially migrate up an adjacent primary dendrite towards the dentate molecular layer. In doing so, dendritic branches on this dendrite become a new primary dendrite. We also propose that this migration underlies the dispersion of the granule cell layer, which was previously suggested by other laboratories; however, the utility of organotypic explant cultures made from Thy1-YFP mice allowed us for the first time to observe fully differentiated granule cell migration and their contribution to the dispersion of the granule cell layer. Granule cell dispersion and distortions to granule cell dendritic structure are common pathologies of the epileptic brain. Both phenomena also occur in adult animal models of epilepsy.

Using bi-transgenic Gli-CreERT2+/-;Green fluorescent protein (GFP) reporter+/- mice, data from Chapter 3 suggests that the pool of hippocampal subventricular zone and/or subgranular cell layer neural progenitor cells active several days prior to a prolonged seizure become disrupted following early-life seizure activity. Specifically, fewer cells within the dentate gyrus were labeled with GFP, and associated with decreased numbers of progenitor cells, immature and mature granule cells. Additionally, we are the first laboratory to show that early-life seizures disrupt the integration of granule cells, which has been a ‘hot’ topic in studies using adult models of epilepsy. In Chapter 4, we used the same bi-transgenic mice, but in an adult model of epilepsy. From this study, we found that the pool of subgranular zone progenitor cells producing progeny following seizures become disrupted, and at later time points, give birth to fewer new granule cells. Additionally, this study implicates that granule cells born during the first week following a seizure may exhibit an accelerated rate of maturation. The results of these studies will hopefully spur further research into the underlying cellular signaling pathways involved in the formation of basal dendrites on dentate granule cells, etcopic localization of granule cells to the hilus and dentate molecular layers and granule cell layer dispersion. Additionally, future studies of cell-fate mapping in the early-life seizure and adult model of epilepsy should include the use of alternate tamoxifen injection time points to determine if progenitors active before or after an epileptogenic stimulus respond differently than the conditions tested here.

Committee:

Steve Danzer, PhD (Committee Chair); Neil Richtand, MD (Committee Member); Kim Seroogy, PhD (Committee Member); Kenneth Campbell, PhD (Committee Member); Katherine Holland-Bouley, MD, PhD (Committee Member)

Subjects:

Neurology

Keywords:

plasticity;dentate granule cell;recurrent basal dendrite;epilepsy;neurogenesis;early-life seizure

Mehta, Manish P.Prediction of manufacturing operations sequence using recurrent neural networks
Master of Science (MS), Ohio University, 1997, Industrial and Manufacturing Systems Engineering (Engineering)
Prediction of manufacturing operations sequence using recurrent neural networks

Committee:

Luis Rabelo (Advisor)

Subjects:

Engineering, Industrial

Keywords:

Recurrent Neural Networks; Computer-Aided Process Planning; Automation and Computer Integrated Manufacturing

Vigraham, Saranyan A.An Analog Evolvable Hardware Device for Active Control
Doctor of Philosophy (PhD), Wright State University, 2007, Computer Science and Engineering PhD
The field of Evolvable Hardware (EH) has recently gained a lot of interest due to the novel methodology it offers for designing electrical circuits and machines. EH techniques involve configuring a reconfigurable hardware platform with the aid of learning engines such as evolutionary algorithms. The EH devices normally act as closed loop controllers with the capability of learning necessary control laws adaptively. Current EH practices have several shortcomings, which have restricted their use as reliable controllers. This dissertation will present an improved EH device based on behavioral reconfigurability that addresses the current open challenges in the field of analog Evolvable Hardware. This EH device is based on Continuous Time Recurrent Neural Network (CTRNN). The design and implementation of the CTRNN-EH device and a custom designed evolutionary learning engine will be presented in this work. In addition to answering the open challenges in the field of EH, this dissertation will also provide a novel programming circuitry to by which a VLSI CTRNN can be effectively programmed. Furthermore, a closed loop calibration scheme based on Evolutionary Algorithms is presented to address the effects of random offset variations in the CTRNN design.

Committee:

John Gallagher (Advisor)

Subjects:

Computer Science

Keywords:

Analog VLSI; Evolvable Hardware; Continuous Time Recurrent Neural Network; Evolutionary Algorithms; Neural Networks in hardware

Obeidat, Ahmed ZayedNew Insights into the Spinal Recurrent Inhibitory Pathway Normally and After Motoneuron Regeneration
Doctor of Philosophy (PhD), Wright State University, 2013, Biomedical Sciences PhD
Despite more than seven decades of intensive research, uncertainty is the hallmark of spinal recurrent inhibition. The simplest possible structure that is formed between the a-motoneuron and its inhibitory interneurons has been the subject of long lasting scientific debate. To date, there is no consensus on the functional significance of this circuit. Even the simplest assumption of a negative feedback loop does not hold true. The current work used the technique of in vivo intracellular recording from the adult rat a-motoneurons to study the normal function and the plasticity after nerve injury and regeneration of this simple, yet intricate spinal circuit. The long lasting notion that inhibition must adversely affect neuronal firing rates has been challenged and the counter-intuitive finding that recurrent inhibition can increase firing rate under certain circumstances is reported for the first time. In addition, recurrent inhibition was found to strongly affect action potential spike timing and was found to prolong the duration of repetitive firing of a-motoneurons. Furthermore, the circuit behavior at different frequencies has been examined and novel findings are reported. The circuit adaptation to peripheral nerve injury and successful regeneration was studied. Results showed that peripheral nerve regeneration failed to restore the structure and function of this central circuit. In conclusion, the current thesis calls for a reevaluation of the concept that recurrent inhibition must suppress a-motoneuron firing and suggests that inhibition in general plays more of a role in modulating firing behavior. Finally, another example of permanent central nervous system dysfunction despite successful peripheral recovery is reported and perhaps adds to the permanent functional deficits that remain in victims of peripheral nerve injury.

Committee:

Timothy Cope, Ph.D. (Advisor); Francisco Alvarez Leefmans, M.D., Ph.D. (Committee Member); Steven Berberich, Ph.D. (Committee Member); Sherif Elbasiouny, Ph.D., PE, PEng. (Committee Member); David Ladle, Ph.D. (Committee Member); Mark Rich, M.D., Ph.D. (Committee Member)

Subjects:

Biomedical Research; Neurology; Neurosciences

Keywords:

Renshaw cell; recurrent inhibition; motoneuron; firing rate; peripheral nerve injury; frequency dynamics; circuit adaptation; spinal cord