MKMichael N. Kozicki

Michael N. Kozicki

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Michael N. Kozicki has a total of 78 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Value of Information of Improved Traceability in Fresh Produce Markets

    by Kelsey Vourazeris, Mark R. Manfredo, Michael N. Kozicki

    Traceability plays an important role in promoting a safe food supply by fostering transparent information exchange along the food supply chain. New technological innovations have the potential to improve traceability outcomes, as greater transparency along the food supply chain can aid in pinpointing precise origins of the contamination, leading to faster and more focused recalls. For instance, a robust technology‐enhanced traceability system could have helped mediate the economic damage experienced in the 2006 spinach market due to an E. coli outbreak, when sales volume declined rapidly, consumer confidence decreased, and prices dropped dramatically due to this shock. However, the adoption of new and innovative traceability technologies may face some resistance due to a lack of proper evaluation of their financial value, especially given the difficulty conceptualizing positive cash flows from such an investment. In this paper we attempt to shed light on this issue by estimating the value of information (VOI) from improved traceability in the leafy greens industry. Specifically, we model the problem of a representative romaine lettuce farm that faces declines in revenue due to exogenous food safety shocks impacting market‐level sales volumes and prices. Using data from the CDC's National Outbreak Reporting System, we employ a Markov chain approach to estimate steady‐state probabilities that reflect the dynamics of food safety incidents related to romaine lettuce over time. Under the proposition that enhanced traceability should reduce both the duration and severity of a food safety incident, we modify these steady‐state probabilities to simulate these improvements. Therefore, the VOI measure considers the value of system‐wide traceability improvement realized at a representative farm level by examining the expected present value of farm revenue under states of improved traceability relative to the expected present value of farm revenue considering the original steady‐state probabilities. Our results indicate that system‐wide improvements in traceability do provide value at the enterprise level, especially when a food safety incident creates a negative shock to market prices and volumes sold.

  • Open Access

    On the impact of pre-training datasets for matching dendritic identifiers using residual nets

    by Baaz Jhaj, Ankita Shukla, Pavan Turaga, Michael Kozicki

    Dendrites are easy to synthesize branching structures that exhibit randomness; yet they are unique, non-repeatable, and identifiable with the right algorithmic innovations. This has created a novel application area where manufactured dendritic structures are being used as product identifiers - essentially "fingerprints for things". Unlike barcodes, which are linear structures, dendrites exhibit spatial randomness. This, coupled with a unique optical signal generated by light scattering from material inhomogeneities, ensures that each dendrite is unique and unclonable. While there have not yet been any established methods on reading dendritic patterns for verification using image data, identifying dendrites using computer vision techniques could have high potential. Due to limited data and low variance, dendrite identification can be considered to be a fine-grained classification task. In this paper, we examine how the selection of pre-trained models influences dendrite classification. The dendrites we work with share similarity to human fingerprints, thus we begin with a model trained for matching fingerprint data to extract features relevant to dendrites. Additionally, we explore broader pre-training approaches, using ImageNet-1K for our second model and ImageNet-21K for our third model. Surprisingly, our results indicate that even with the visual similarity with human fingerprints, more general pre-training with common image datasets achieves better performance on dendrite classification.

  • Impact of TID on the Analog Conductance and Training Accuracy of CBRAM-Based Neural Accelerator

    by P. Apsangi, N. Chamele, H. J. Barnaby, Y. Gonzalez-Velo, K. E. Holbert, M. N. Kozicki

    The changes caused by total ionizing dose (TID) in the conductance of the analog response of Ag-Ge 30 Se 70 conductive bridge random access memory (CBRAM) based synapses are studied. The conductance was seen to be severely affected by 60 Co gamma-ray exposure. The devices were tested up to a TID of 1 Mrad(Ge 30 Se 70 ). Scanning electron microscopy imaging supports the conclusion that the device degradation is due to loss of filament contact after irradiation. The effect of irradiation was further analyzed by simulating the devices in an artificial neural network. The training accuracy was seen to degrade from 85% to 15% with increasing TID levels.


  • Materials Characterization and Electrical Performance of Bilayer Structures for Enhanced Electrodeposition in Programmable Metallization Cells

    by Ninad Chamele, Mehmet B. Balaban, Arshey Patadia, Smitha S. Swain, Yago Gonzalez‐Velo, Manuel Roldan, Michael Kozicki

    Long channel lateral programmable metallization cell (PMC) devices, in which a metal electrodeposit is formed on a solid electrolyte between two coplanar electrodes several microns apart, have a variety of applications beyond standard microelectronics. Whereas fast electrodeposit growth over considerable distances can be achieved using high ion mobility chalcogenide electrolytes, the use of more foundry‐friendly materials, including copper‐doped transition metal oxides, leads to extremely low electrodeposit growth rates due to their low ion mobility and high resistivity. In this paper, a material system comprising a Cu2O/CuWO3 bilayer formed by low temperature oxidation of Cu on WO3 is examined. The Cu2O overlayer provides a low resistance parallel path for electrons, allowing electrodeposition to occur along the entire length of the device, thereby increasing the rate at which the electrodeposit fills the channel. A study of devices prepared under different conditions, resulting in different copper oxide thicknesses and copper concentrations in the layers, shows the influence of processing conditions on material composition, activation energy for current transport, electrodeposit morphology, and electrical characteristics. It is demonstrated that a very wide range of operating conditions is possible, including the bridging of 14 μm long channels in under 3 s using 2 V bias.


  • Open Access

    The viability of analog-based accelerators for neuromorphic computing: a survey

    by Mirembe Musisi-Nkambwe, Sahra Afshari, Hugh Barnaby, Michael N Kozicki, Ivan Sanchez Esqueda

    Focus in deep neural network hardware research for reducing latencies of memory fetches has steered in the direction of analog-based artificial neural networks (ANN). The promise of decreased latencies, increased computational parallelism, and higher storage densities with crossbar non-volatile memory (NVM) based in-memory-computing/processing-in-memory techniques is not without its caveats. This paper surveys this rich landscape and highlights the advantages and challenges of emerging NVMs as multi-level synaptic emulators in various neural network types and applications. Current and potential methods for reliably programming these devices in a crossbar matrix are discussed, as well as techniques for reliably integrating and propagating matrix products to emulate the well-known MAC-like operations throughout the neural network. This paper complements previous surveys, but most importantly uncovers further areas of ongoing research relating to the viability of analog-based ANN implementations based on state-of-the-art NVM technologies in the context of hardware accelerators. While many previous reviews of analog-based ANN focus on device characteristics, this review presents the perspective of crossbar arrays, peripheral circuitry and the required architectural and system considerations for an emerging memory crossbar neural network.


  • Open Access

    Information in electrodeposited dendrites

    by Michael N. Kozicki

    Dendrites are structures that develop with a continuously branching tree-like form. Such patterns are found in many aspects of the natural world, which indicates the universality of their topology. This review presents an examination of dendritic structures, addressing their stochasticity and fractal character, and exploring their information content or more specifically their ability to provide a very large number of unique patterns that may be used as a novel form of item identification. A brief summary of fractals and their dimensionality is presented and applied to the well-known diffusion limited aggregate (DLA) dendritic construct. Dendrites formed by electrochemical ‘self-assembly’ are explored and examples given of their formation under different conditions. Stochastic variations in the self-similar Y-shaped symbol that underlies these fractals can carry information, leading to significant entropy, even though the structural entropy of the overall pattern is relatively small.


  • Open Access

    Characterization of a novel, low-cost, scalable ozone gas system for sterilization of N95 respirators and other COVID-19 related use cases.

    by Nikhil Dave, Katie Sue Pascavis, John M Patterson, Michael Kozicki, David W Wallace, Abhik Chowdhury, Morteza Abbaszadegan, Absar Alum, Pierre Herckes, Zhaobo Zhang, et al.

    Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), an elusive and highly pathogenic agent, has resulted in the ongoing COVID-19 pandemic affecting numerous populations worldwide. New studies investigating the tenacity of SARS-CoV-2 have highlighted its ability to persist on a myriad of surfaces for several days, including gowns and shoes. As a result, there is a global need for sterilization of a variety of potentially-contaminated items, ranging from clothing to personal protective equipment like face coverings. To this end, we have designed and constructed a cost-effective, scalable, and sustainable sterilization system that uses ozone gas to inactivate viral particles. We sought to determine the efficacy of the system in the sterilization of viral particles as well as its ability to sterilize N95 respirators for reuse. N95 respirators inoculated with P22 bacteriophage and sterilized in the ozone system showed a 6-log 10 reduction in viral load when treated at 25 ppm for 150 minutes. Further, N95 respirators treated with five 150-minute cycles at 35 ppm for a total concentration-time product (CT) of 26,250 ppm min in the ozone system showed comparable filtration efficiency to untreated N95 respirators in a 50 to 200 nmr particulate challenge filtration test. Interestingly, the surgical N95 respirators tested showed complete inactivation of fluid resistance and degradation of the elasticity of polyisoprene straps after five cycles in the sterilization system. Taken together, these data suggest that while our ozone system may negatively affect certain protective aspects of surgical N95 respirators, it does effectively sterilize viral particles and can be utilized for a multitude of other use cases, including sterilizing polypropylene face coverings after potential SARS-CoV-2 contamination. In addition to providing long-term environmental benefits, deployment of this system during the ongoing pandemic reduces the risk of COVID-19 community transmission while conserving monetary resources otherwise spent on the continuous purchase of disposable face coverings.


  • Open Access

    Characterization of a novel, low-cost, scalable vaporized hydrogen peroxide system for sterilization of N95 respirators and other COVID-19 related personal protective equipment

    by Nikhil Dave, Katie Sue Pascavis, John M Patterson, David W Wallace, Abhik Chowdhury, Morteza Abbaszadegan, Absar Alum, Pierre Herckes, Zhaobo Zhang, Michael Kozicki, et al.

    Due to the virulence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the pathogen responsible for the respiratory disease termed COVID-19, there has been a significant increase in demand for surgical masks and N95 respirators in medical clinics as well as within communities operating during the COVID-19 epidemic. Thus, community members, business owners, and even medical personnel have resorted to alternative methods for sterilizing face coverings and N95 respirators for reuse. While significant work has shown that vaporized hydrogen peroxide (VHP) can be used to sterilize N95 respirators, the cost and installation time for these sterilization systems limit their accessibility. To this end, we have designed and constructed a novel, cost-effective, and scalable VHP system that can be used to sterilize N95 respirators and other face coverings for clinical and community applications. N95 respirators inoculated with P22 bacteriophage showed a greater than 6-log 10 reduction in viral load when sterilized in the VHP system for one 60-minute cycle. Further, N95 respirators treated with 20 cycles in this VHP system showed comparable filtration efficiency to untreated N95 respirators in a 50 to 200 nanometer particulate challenge filtration test. While a 23% average increase in water droplet roll-off time was observed for N95 respirators treated with 5 cycles in the sterilization, no breakdown in fluid resistance was detected. These data suggest that our VHP system is effective in sterilizing N95 respirators and other polypropylene masks for reuse. Relating to the present epidemic, deployment of this system reduces the risk of COVID-19 community transmission while conserving monetary resources otherwise spent on the continuous purchase of disposable N95 respirators and other face coverings. In summary, this novel, scientifically validated sterilization system can be easily built at a low cost and implemented in a wide range of settings.


  • Structure and charge transport of amorphous Cu-doped Ta2O5 : An ab initio study

    by Rajendra Thapa, Bishal Bhattarai, M. N. Kozicki, Kashi N. Subedi, D. A. Drabold

    In this paper, we present ab initio computer models of Cu-doped amorphous Ta2O5, a promising candidate for conducting bridge random access memory memory devices, and study the structural, electronic, charge transport, and vibrational properties based on plane-wave density-functional methods. We offer an atomistic picture of the process of phase segregation/separation between Cu and Ta2O5 subnetworks. Electronic calculations show that the models are conducting with extended Kohn-Sham orbitals around the Fermi level. In addition to that, we also characterize the electronic transport using the Kubo-Greenwood formula modified suitably to calculate the space-projected conductivity (SPC) [K. Prasai, K. N. Subedi, K. Ferris, P. Biswas, and D. A. Drabold, Phys. Stat. Solidi Rapid Res. Lett. 12, 1800238 (2018)]. Our SPC calculations show that Cu clusters and undercoordinated Ta adjoining the Cu are the conduction-active parts of the network. We also report information about the dependence of the electrical conductivity on the connectivity of the Cu submatrix. Vibrational calculations for one of the models has been undertaken with an emphasis on localization and animation of representative modes.


  • Open Access

    Structure and charge transport of amorphous $Cu$-doped $Ta_2O_5$ : An ab initio study

    by Rajendra Thapa, Bishal Bhattarai, M. N. Kozicki, Kashi N. Subedi, D. A. Drabold

    In this paper, we present ab initio computer models of Cu-doped amorphous Ta2O5 , a promising candidate for Conducting Bridge Random Access Memory (CBRAM) memory devices, and study the structural, electronic, charge transport and vibrational properties based on plane-wave density functional methods. We offer an atomistic picture of the process of phase segregation/separation between Cu and Ta2O5 subnetworks. Electronic calculations show that the models are conducting with extended Kohn-Sham orbitals around the Fermi level. In addition to that, we also characterize the electronic transport using the Kubo-Greenwood formula modified suitably to calculate the space-projected conductivity (SPC). Our SPC calculations show that Cu clusters and under-coordinated Ta adjoining the Cu are the conduction-active parts of the network. We also report information about the dependence of the electrical conductivity on the connectivity of the Cu sub-matrix. Vibrational calculations for one of the models has been undertaken with an emphasis on localization and animation of representative modes.