Progress 09/01/23 to 08/31/24
Outputs
Target Audience:Goal 1: Data exploration, curating and preliminary analysis Detailed analysis of AMS Movement and AMS Terminal Market data were conducted with the objective of constructing realistic commodity flow networks for tomatoes. Data from Terminal Market contains daily/weekly tomato price and origin information for select wholesale terminal markets. These were analyzed in detail to with respect to Mature Greens, Immature Greens, Vine Ripe, cherry tomatoes, and plum-type tomatoes. Movement Report Data from USDA AMS contains weekly shipment by weight and origin. The origins include California Central, Alabama, Arkansas, Florida, Michigan, North Carolina, South Carolina, Tennessee, and Virginia. The data also includes external origins in the form of imports from Canada, the Dominican Republic, Guatemala, Honduras, and Mexico. The data was collected from 2021 to 2024, only required columns chosen, and standardized. Data from AgCensus and various remote sensing-based datasets were explored to map farms and other agricultural infrastructure. Goal 2: Modeling and analysis of commodity flows. The two datasets were linked by connecting the sources from Terminal Market data and regions from the Movement data. In some cases, two sources in the former may have to be mapped to one region in the latter. These decisions were taken based on geographic proximity considerations. While the Terminal market data provides edge attributes, the Movement data provides node attributes. Given a terminal node, a source node, their features (such as population, production, and consumer price index), time, and source-terminal attributes such as distance, the objective is to determine if there is commodity flow from source to terminal. There are only a subset of terminals for which commodity flow information is available. Hence, this is a semi-supervised learning formulation where the objective is to be able to predict the flows for the unobserved terminals. Firstly, we would like to leverage the spatial location information of the various terminals and the sources through the use of a distance graph that can account for the spatial correlations. The idea is that terminals that are close by are probably sourcing the commodity in similar ways as opposed to those that are far from each other. Similar relations hold for sources as well. Secondly, we would like to leverage seasonal or temporal information as the flow of agricultural commodities are driven by their production that depends on the time of the year as well as the location. The approach to construct this network is a product graph based graph neural network. Multiple semi-supervised methods are being applied as baselines to compare with the proposed approach. Remote-sensing and deep learning methods were applied to map agricultural infrastructure. These methods and the data products will be applied to estimate accurately, tomato and other vegetable production in the network models. Goal 3(b) Analyze spread under various hypothetical scenarios The work from 2021 on interventions was finalized and submitted to Nature Computational Science. Additional experiments were conducted regarding robustness to sampling complexity and model uncertainty. A summary of the work is provided. Optimal control of spread processes over networks is a challenging problem even for simple diffusion models. Biological invasions are characterized by multiple spread pathways and time-varying network attributes. In this setting, we study the problem of region-wide interventions, where the objective is to find an optimal set of regions represented by groups of nodes in the underlying network to minimize the spread given budget constraints, intervention delays, and a spread scenario. We present an approach applicable to a general class of diffusion models based on integer linear programming and sample average approximation, and prove rigorous bounds on its performance. We apply this method to the spread of a representative agricultural pest. Our approach provides near-optimal solutions and consistently outperforms considered baselines. Our results highlight the importance of scenario-specific control and suggest that early intervention has the benefit of a significant reduction in spread under low budget as well as stable solutions under model uncertainty. Publications: A High-Resolution, US-scale Digital Similar of Interacting Livestock, Wild Birds, and Human Ecosystems with Applications to Multi-host Epidemic Spread. Abhijin Adiga, Ayush Chopra, Mandy L Wilson, SS Ravi, Dawen Xie, Samarth Swarup, Bryan Lewis, John Barnes, Ramesh Raskar and Madhav V Marathe. Submitted to PNAS. Scenario-specific Control of Multi-pathway Spread Processes: Application to Biological Invasions, Prathyush Sambaturu, Manisha Sudhir, Hongze Chen, Anil Vullikanti, Rangaswamy Muniappan, and Abhijin Adiga. Submitted to Nature Computational Science. ? Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?One postdoctoral fellow, two postgraduate students, and one undergraduate were mentored. One postgraduate student worked on surveillance problems in the context of spreading processes over networks. Another postgraduate student and the postrdoctoral fellow worked on the commodity flow problem and the graph neural network-based approach. The postgraduate student also assisted in experimentation for the interventions paper (coauthor of the submitted paper). The undergraduate student was tasked with analyzing various data sources (Goals 1 and 2). He also worked on developing deep learning surrogates for spread processes over networks. ? How have the results been disseminated to communities of interest?Papers have been submitted for possible publication. What do you plan to do during the next reporting period to accomplish the goals?Paper on constructing commodity flow networks will be written up and submitted. Based on reviews, work on the pending interventions paper will be completed. Datasets will be organized and made available in a data repository or github. The multipathway simulator along with the networks have already been made public. This repository will be updated. ?
Impacts
What was accomplished under these goals?
Goal 1: Data exploration, curating and preliminary analysis Detailed analysis of AMS Movement and AMS Terminal Market data were conducted with the objective of constructing realistic commodity flow networks for tomatoes. Data from Terminal Market contains daily/weekly tomato price and origin information for select wholesale terminal markets. These were analyzed in detail to with respect to Mature Greens, Immature Greens, Vine Ripe, cherry tomatoes, and plum-type tomatoes. Movement Report Data from USDA AMS contains weekly shipment by weight and origin. The origins include California Central, Alabama, Arkansas, Florida, Michigan, North Carolina, South Carolina, Tennessee, and Virginia. The data also includes external origins in the form of imports from Canada, the Dominican Republic, Guatemala, Honduras, and Mexico. The data was collected from 2021 to 2024, only required columns chosen, and standardized. Data from AgCensus and various remote sensing-based datasets were explored to map farms and other agricultural infrastructure. Goal 2: Modeling and analysis of commodity flows. The two datasets were linked by connecting the sources from Terminal Market data and regions from the Movement data. In some cases, two sources in the former may have to be mapped to one region in the latter. These decisions were taken based on geographic proximity considerations. While the Terminal market data provides edge attributes, the Movement data provides node attributes. Given a terminal node, a source node, their features (such as population, production, and consumer price index), time, and source-terminal attributes such as distance, the objective is to determine if there is commodity flow from source to terminal. There are only a subset of terminals for which commodity flow information is available. Hence, this is a semi-supervised learning formulation where the objective is to be able to predict the flows for the unobserved terminals. Firstly, we would like to leverage the spatial location information of the various terminals and the sources through the use of a distance graph that can account for the spatial correlations. The idea is that terminals that are close by are probably sourcing the commodity in similar ways as opposed to those that are far from each other. Similar relations hold for sources as well. Secondly, we would like to leverage seasonal or temporal information as the flow of agricultural commodities are driven by their production that depends on the time of the year as well as the location. The approach to construct this network is a product graph based graph neural network. Multiple semi-supervised methods are being applied as baselines to compare with the proposed approach. Remote-sensing and deep learning methods were applied to map agricultural infrastructure. These methods and the data products will be applied to estimate accurately, tomato and other vegetable production in the network models. Goal 3(b) Analyze spread under various hypothetical scenarios The work from 2021 on interventions was finalized and submitted to Nature Computational Science. Additional experiments were conducted regarding robustness to sampling complexity and model uncertainty. A summary of the work is provided. Optimal control of spread processes over networks is a challenging problem even for simple diffusion models. Biological invasions are characterized by multiple spread pathways and time-varying network attributes. In this setting, we study the problem of region-wide interventions, where the objective is to find an optimal set of regions represented by groups of nodes in the underlying network to minimize the spread given budget constraints, intervention delays, and a spread scenario. We present an approach applicable to a general class of diffusion models based on integer linear programming and sample average approximation, and prove rigorous bounds on its performance. We apply this method to the spread of a representative agricultural pest. Our approach provides near-optimal solutions and consistently outperforms considered baselines. Our results highlight the importance of scenario-specific control and suggest that early intervention has the benefit of a significant reduction in spread under low budget as well as stable solutions under model uncertainty. Publications: A High-Resolution, US-scale Digital Similar of Interacting Livestock, Wild Birds, and Human Ecosystems with Applications to Multi-host Epidemic Spread. Abhijin Adiga, Ayush Chopra, Mandy L Wilson, SS Ravi, Dawen Xie, Samarth Swarup, Bryan Lewis, John Barnes, Ramesh Raskar and Madhav V Marathe. Submitted to PNAS. Scenario-specific Control of Multi-pathway Spread Processes: Application to Biological Invasions, Prathyush Sambaturu, Manisha Sudhir, Hongze Chen, Anil Vullikanti, Rangaswamy Muniappan, and Abhijin Adiga. Submitted to Nature Computational Science. ?
Publications
|
Progress 09/01/22 to 08/31/23
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?One graduate student and one undergraduate student were mentored. The graduate student worked on surveillance problems in the context of spreading processes over networks. The undergraduate student was tasked with analyzing various data sources (Goals 1 and 2). He also worked on developing deep learning surrogates for spread processes over networks. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Task 2: Modeling and analysis of agricultural commodity flows Machine learning based methods are being applied to infer commodity flow links between locations. Task 3: Analyze spread under various hypothetical scenarios The focus will be on spatial surveillance algorithms. Novel Steiner-tree-based algorithms are being developed for directed weighted temporal networks. Multiple publications planned for this period: 1. Controlling the multi-pathway spread of invasive species using multi-scale network interventions. Journal version (Manisha Sudhir's thesis work) 2. Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems. Journal version (revised version being submitted) 3. Reconstructing an Epidemic Outbreak Using Steiner Connectivity (extension of Mishra et al. 2023 proceedings paper)
Impacts
What was accomplished under these goals?
Goal 3(b) Analyze spread under various hypothetical scenarios Mishra et al. 2022: We develop a community detection method to understand interdependencies in a food network with respect to spread processes. The objective is to find groups of nodes in the network (such as counties or states) with strong influence within groups. This will aid in the design of effective surveillance and control methods. Community detection in networks is extensively studied from a structural perspective, but very few works characterize communities with respect to dynamics on networks. We propose a generic framework based on Moore-Shannon network reliability for defining and discovering communities with respect to a variety of dynamical processes. This approach extracts communities in directed edge-weighted networks which satisfy strong connectivity properties as well as strong mutual influence between pairs of nodes through the dynamical process. We apply this framework to global as well as national-level vegetable and cereal networks. We compare our results with modularity-based approach, and analyze community structure across commodities, evolution over time, and with regard to dynamical system properties. Mishra et al. 2023: Only a subset of infections/invasions is actually observed in an outbreak, due to multiple reasons such as under-reporting or being unable to identify the invading organism. Therefore, reconstructing an epidemic cascade given some observed cases is an important step in responding to such an outbreak. A maximum likelihood solution to this problem (referred to as CASCADEMLE) can be shown to be a variation of the classical Steiner subgraph problem, which connects a subset of observed infections. In contrast to prior works on epidemic reconstruction, which consider the standard Steiner tree objective, we show that a solution to CASCADEMLE, based on the actual MLE objective, has a very different structure. We design a logarithmic approximation algorithm for CASCADEMLE, and evaluate it on multiple synthetic and social contact networks, including a contact network constructed for a hospital. Our algorithm has significantly better performance compared to a prior baseline. Harrison et al. 2023: We consider the setting of cascades that result from contagion dynamics on large realistic networks. We address the question of whether the structural properties of a (partially) observed cascade can characterize the contagion scenario and identify the interventions that might be in effect. Using epidemic spread as a concrete example, we study how social interventions such as compliance in social distancing, extent (and efficacy) of vaccination, and the transmissibility of disease can be inferred. The techniques developed are more generally applicable to other contagions as well. Through a machine learning approach, coupled with parameter significance analysis, our experimental results show that subgraph counts of the graph induced by the cascade can be used effectively to characterize the contagion scenario even during the initial stages of the epidemic. Further, we show that our approach performs well even for partially observed cascades. These results demonstrate that cascade data collected from delimiting surveys or contact tracing can provide valuable information about the contagion scenario. Publications: Mishra, Ritwick, et al. "Community Detection Using Moore-Shannon Network Reliability: Application to Food Networks." International Conference on Complex Networks and Their Applications. Cham: Springer International Publishing, 2022.V Mishra, Ritwick, et al. "Reconstructing an Epidemic Outbreak Using Steiner Connectivity." Proceedings of the AAAI Conference on Artificial Intelligence. Vol. 37. No. 10. 2023. Harrison, Galen, et al. "Identifying Complicated Contagion Scenarios from Cascade Data." Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 2023.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2022
Citation:
Mishra, Ritwick, et al. "Community Detection Using Moore-Shannon Network
Reliability: Application to Food Networks." International Conference on
Complex Networks and Their Applications. Cham: Springer International
Publishing, 2022.V
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2023
Citation:
Mishra, Ritwick, et al. "Reconstructing an Epidemic Outbreak Using Steiner
Connectivity." Proceedings of the AAAI Conference on Artificial
Intelligence. Vol. 37. No. 10. 2023.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2023
Citation:
Harrison, Galen, et al. "Identifying Complicated Contagion Scenarios from
Cascade Data." Proceedings of the 29th ACM SIGKDD Conference on Knowledge
Discovery and Data Mining. 2023.
|
Progress 09/01/21 to 08/31/22
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Two undergraduate students, one graduate student, one postdoc, and a high school student were mentored during this period. One undergraduate student was tasked with analyzing various data sources (Goals 1 and 2). In this process the student was primarily mentored in machine learning techniques and network science. Another undergraduate student was tasked with network analysis. The students also presented research papers related to network diffusion processes. The graduate student has been working on two directions. The first direction corresponds to cascade reconstruction and surveillance type problems. The second work corresponds to dynamics-based community detection that resulted in a conference paper and a workshop presentation (presented by the student). The high school student has worked on the dynamics-based community detection and helped collect data on economic impact. The postdoc contributed to the application of artificial neural networks for the problem of network construction. How have the results been disseminated to communities of interest?Results have been communicaed through publications. What do you plan to do during the next reporting period to accomplish the goals?Based on our analysis, there is a need to obtain sample data on end-to-end tomato trade between locations. Current effort is on obtaining such data. Even information providing evidence of flow (presence or absence of a link) can be used for learning and validation of networks. Several new features are being added to the multi-pathway simulator. These include several kernels for short-distance spread such as Gaussian, exponential, Weibull, etc. Also, location-specific population growth parameter is being implemented. This is particularly important when simulating for the conternimus USA, where the growth rates of T. absoluta can differ greatly. New surveillance and control algorithms are being developed using state-of-the-art algorithmic approaches. Economic impact studies will be conducted using the data collected, spread model and the partial equilibrium approach. Multiple publications planned for this period: 1. Controlling the multi-pathway spread of invasive species using multi-scale network interventions. Journal version (Manisha Sudhir's thesis work) 2. Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems. Journal version (extension of conference paper in Complex Networks and Applications) 3. Assesing the pathways of introduction and spread of T. absoluta in North America. 4. Reconstructing an Epidemic Outbreak Using Steiner Connectivity
Impacts
What was accomplished under these goals?
Goal 1: Data exploration, curating and preliminary analysis More datasets will be explored. This includes at least the following sources: (i) USDA APHIS Pest interception database to which we recently gained access, (ii) more data for economic impact analysis from ERS (iii) import-export data, and (iii) operations data from the industry collaborator. Goal 2: Modeling and analysis of commodity flows. We studied community structures in commodity flow networks. Very few works characterize communities with respect to diffusion processes on networks. Understanding the community structure of international food networks particularly in the context of spread processes representing cascading failures and biological invasions can help inform surveillance and control strategies. We propose a generic framework based on Moore-Shannon network reliability for defining and discovering communities with respect to a variety of dynamical processes. This approach extracts communities in directed edge-weighted networks which satisfy strong connectivity properties as well as strong mutual influence between pairs of nodes through the dynamical process. We apply this framework to country-to-country networks from FAO and US domestic flow networks from FAF. We compare our results with modularity- based approach, and analyze community structure across commodities, evolution over time, and with regard to dynamical system properties. We have considered motif counting in temporal networks to analyze the vulnerability in commodity flow networks arising due to seasonality in production. Motifs are small template subgraphs in the network with special properties. The number of motifs of a particular kind can indicate anomalies in the structure that could indicate vulnerability to invasion or rapid spread. Goal 3: Multi-pathway models for invasive species spread. The network-based modeling and simulation approaches applied to study biological invasions tend to be largely domain-specific, lacking any graph theoretic formalisms, and do not take advantage of more recent developments in network science. To rigorously understand these processes, we developed a generic multi-scale spatial network framework that is applicable to a wide range of models developed in the literature on biological invasions. A key question we address is the following: how do individual pathways and their combinations influence the rate and pattern of spread? The analytical complexity arises more from the multi-scale nature and complex functional components of the networks rather than from the sizes of the networks. We present theoretical bounds on the spectral radius and the diameter of multi-scale networks. These two structural graph parameters have established connections to diffusion processes. Specifically, we study how network properties, such as spectral radius and diameter are influenced by model parameters. Further, we analyze a multi-pathway diffusion model from the literature by conducting simulations on synthetic and real-world networks and then use regression tree analysis to identify the important network and diffusion model parameters that influence the dynamics. Publications: 1. Mishra et al., Community Detection using Moore-Shannon Network Reliability: Application to Food Networks, International Conference on Complex Networks and Their Applications 2022. 2. Mishra et al., Communities in Directed Weighted Food Networks using Moore-Shannon Network Reliability, Workshop "Communities in networks" in NetSci 2022. 3. Adiga et al., Network Models and Simulation Analytics for Multi-scale Dynamics of Biological Invasions, Frontiers BigData 4. Adiga et al., Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems, International Conference on Complex Networks and Their Applications 2021. 5. Manisha Sudhir. Controlling Diffusion on Multi-Pathway Spatial Networks: Application to Biological Invasions, Masters thesis, UVA.
Publications
- Type:
Journal Articles
Status:
Accepted
Year Published:
2022
Citation:
Adiga, A., Palmer, N., Baek, Y. Y., Mortveit, H., & Ravi, S. S. (2022). Network Models and Simulation Analytics for Multi-scale Dynamics of Biological Invasions. Frontiers in big Data, 5.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2021
Citation:
Adiga, A., Palmer, N., Sinha, S., Waghalter, P., Dave, A., Lazarte, D. P., ... & Marathe, M. (2021, November). Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems. In International Conference on Complex Networks and Their Applications (pp. 168-179). Springer, Cham.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2022
Citation:
Mishra et al., Community Detection using Moore-Shannon Network
Reliability: Application to Food Networks, International Conference on
Complex Networks and Their Applications 2022.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2022
Citation:
Mishra et al., Communities in Directed Weighted Food Networks using
Moore-Shannon Network Reliability, Workshop "Communities in networks" in
NetSci 2022.
- Type:
Theses/Dissertations
Status:
Accepted
Year Published:
2021
Citation:
Manisha Sudhir. Controlling Diffusion on Multi-Pathway Spatial Networks:
Application to Biological Invasions, Masters thesis, UVA.
- Type:
Theses/Dissertations
Status:
Accepted
Year Published:
2022
Citation:
Prathyush Sambaturu. Controlling Epidemics on Networks Using Stochastic Optimization Techniques, PhD thesis, UVA.
|
Progress 09/01/20 to 08/31/21
Outputs
Target Audience:
Nothing Reported
Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Three graduate students and one postdoc were mentored during this period. Two graduate students were tasked with analyzing various data sources (Goals 1 and 2). In this process the student was primarily mentored in machine learning techniques and network science. The student was also tasked with presenting research papers related to network diffusion processes. As part of her thesis work, the third graduate student worked on developing intervention algorithms. To this end, she completed an efficient implementation of the multi-pathway simulator using vectorized programming in pandas. She assisted in the design and development of various intervention schemes. The student helped in developing advanced optimization algorithms for interventions. She also conducted extensive experiments on real-world networks to evaluate the algorithms. Thesis has been defended, and a research paper is under review. The postdoc continues to explore the possibility of applying artificial neural networks for the problem of network construction (Goal 1). Data and methodological challenges have been identified. ? How have the results been disseminated to communities of interest?Adiga et al., Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems. Submitted to Complex Networks 2021. Sambaturu et al., Controlling Diffusion on Multi-Pathway Spatial Networks: Application to Biological Invasions, Under review. Eubank et al., Perturbative methods for monotonic probabilistic satisfiability problems, under preparation. What do you plan to do during the next reporting period to accomplish the goals?Goal 1: Data exploration, curating and preliminary analysis More datasets will be explored. This includes at least the following sources: (i) USDA APHIS Pest interception database to which we recently gained access, (ii) more data for economic impact analysis from ERS (iii) import-export data, and (iii) operations data from the industry collaborator. Goal 2: Modeling and analysis of commodity flows. We will continue to work on modeling commodity flows from FAF networks. The main challenges is estimating commodity specific flows from aggregate vegetable flows. Developing new techniques for analyzing temporal weighted directed networks is critical in the analysis. Methods like eigenvector centrality and betweenness centrality measures are well understood for simple unweighted networks. We are developing algorithms where such measures will be adapted for food flow networks. Goal 3: Multi-pathway models for invasive species spread. We initiated the economic impact assessment with a number of meetings with collaborators. Preliminary datasets have been identified. These will be applied to partial equilibrium models. Goal 4: Uncertainty quantification and sensitivity analysis All the methods developed under the three goals will be rigorously analyzed with regard to robustness. We will analyze the sensitivity of the modeled networks to functional and structural perturbations of the modeled networks. Under these perturbations, we will also analyze how forecasts and outcomes of intervention algorithms change. Multiple publications planned for this period: 1. Network Models and Simulation Analytics for Multi-scale Dynamics of Biological Invasions, Frontiers BigData 2. Controlling the multi-pathway spread of invasive species using multi-scale network interventions. Journal version (conference version under review) 3. Dynamics-based clustering and its application to food flows. 4. Realistic Commodity Flow Networks to Assess Vulnerability of Food Systems. Journal version (conference version under review)?
Impacts
What was accomplished under these goals?
Goal 1: Data exploration, curating and preliminary analysis In the previous period, datasets corresponding to Freight Analysis Framework (FAF), vegetable production from CROPSCAPE and trade matrix from FAO were downloaded, standardized, and stored in a PostgreSQL database. Since the data is from multiple sources, we had to standardize the datasets in order to analyze them in combination. In this period, for temporal disaggregation, tomato growing zones were obtained from https://www.tomatofest.com/Tomato_Growing_Zone_Maps_s/164.htm were digitized by manually assigning counties to corresponding zones as well as mapping growing period to months. This was stored in CSV and JSON formats. Data on economic impact regarding production quantities at state-level, yield, supply and demand elasticities have been collected. However, there is a lot of misalignment in time and spatial resolution. We are investigating methods to remedy this problem. Goal 2: Modeling and analysis of commodity flows. We have developed a general framework for constructing the spatiotemporal representation of production, flow, and consumption of agricultural commodities. These data representations are derived by fusing gridded, administrative-level, survey datasets on production, trade, and consumption. Further, they are periodic temporal networks reflecting seasonal variations in production and trade of the crop. We apply this approach to create networks of tomato flow for three regions -- conterminous United States, Senegal, and Nepal. Using statistical methods and network analysis, we gain insights into spatiotemporal dynamics of production and trade. Our results suggest that agricultural systems are increasingly vulnerable to attacks through trade of commodities due to their vicinity to regions of high demand and seasonal variations in production and flows. For the US, we have used data from CROPSCAPE and FAF in this framework. The monthly flows of specific commodities from annual commodity-aggregated flows. In general, spatiotemporal disaggregation of FAF flows is a hard problem. In our case, it is made harder by the fact that we want to infer commodity-specific flows (for e.g., tomato) from FAF networks. Here, the tomato flow is a component of the flow corresponding to SCTG=3 (other agricultural crops). In the current setup, we have come up with a preliminary model to construct such flows. Goal 3: Multi-pathway models for invasive species spread and interventions. More functional relationships have been added to the multi-pathway model developed in our previous work [McNitt et al. 2019]. We have added the functionality of radial spread based on Haversine distance. Interventions at the node-level as well as locality-level have been implemented. Secondly, using vectorized operations in python programming language we have made the simulator faster and therefore scalable to larger networks. Our current implementation is 10-50 times faster compared to the earlier version. We developed a multi-scale intervention framework with the objective of selecting few locations to setup traps or apply interventions such as pesticides in order to delay or stifle the spread of the pest in the event of its introduction. Optimal control of epidemics is a challenging problem even for simple diffusion processes over static networks. We developed this algorithm for the multi-scale epidemiological process on a temporal network in the context of invasive species spread across a landscape. In this setting, we study the problem of group-scale interventions, where the objective is to find an optimal set of regions represented by groups of nodes to minimize the spread under budget constraints and intervention delays. We present an integer linear programming based algorithm for finding effective group-scale interventions and prove rigorous bounds on its performance. We experimentally evaluate it on several real-world networks constructed in McNitt et al. with respect to budget, model uncertainties, introduction scenarios, intervention delays, and rounding schemes. It provides near-optimal solutions and outperforms considered baselines. Also, the performance of group-scale control compares well with the superior but impractical node-scale version of the algorithm. Further, we analyze our solutions for various seeding scenarios and model parameters. Our results indicate that early intervention has the benefit of significant reduction in spread for low budget and stable solutions under model uncertainty. McNitt, J., Chungbaek, Y. Y., Mortveit, H., Marathe, M., Campos, M. R., Desneux, N., ... & Adiga, A. (2019). Assessing the multi-pathway threat from an invasive agricultural pest: Tuta absoluta in Asia. Proceedings of the Royal Society B, 286(1913), 20191159.
Publications
|
Progress 09/01/19 to 08/31/20
Outputs
Target Audience:
Nothing Reported
Changes/Problems:Disruptions due to COVID-19: 1. Personnel effort: Our team has been involved in COVID modeling work since March 2020 to support policy making for the state and the university. In this regard, much of the resources had to be redirected. As a result, other projects such as this one have not received as much attention as planned at the beginning of this period. 2. Student hiring and mentorship: It has been difficult to hire and remotely mentor students. What opportunities for training and professional development has the project provided?Two graduate students and one postdoc were mentored during this period. One graduate student was tasked with downloading and analyzing various data sources (Goals 1 and 2). In this process the student was primarily mentored in database management systems, GIS and network analytics, and developed expertise in python programming language and PostgreSQL. The student was also tasked with presenting research papers related to network construction and related problems in linear and non-linear optimization. The second graduate student has been tasked with developing the invasive species spread simulation model, which is being implemented using python programming language (Goals 2 and 3). The main objectives are two-fold: (i) extend a previously implemented model to make it more general and efficient and (ii) enable integration with intervention algorithms framework when these are developed at a later stage in the project. The student is being mentored in vectorized operations in advanced python packages Numpy and Pandas. The student is also developing advanced optimization algorithms for monitoring and interventions. The student is being mentored on network dynamics and linear programming to this end. The postdoc is exploring the possibility of applying artificial neural networks for the problem of network construction (Goal 1). How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Goal 1: Data exploration, curation and preliminary analysis More datasets will be explored. This includes at least the following sources: (i) USDA APHIS Pest interception database to which we recently gained access (ii) data for economic impact analysis from ERS. Goal 2: Modeling and analysis of commodity flows. We will continue to work on modeling commodity flows from FAF networks. The main challenges are (i) estimating commodity specific flows from aggregate vegetable flows, and (ii) estimating time varying networks representing seasonal commodity flows from annual production and trade data. Literature survey indicates that there is no work done in this area to the best of our knowledge. However, there are several works on disaggregating available commodity flows to finer resolution spatial networks. Goal 3: Multi-pathway models for invasive species spread. The ongoing work on the multi-pathway model to make it efficient and more general will be completed (Goal 3a). The major work during this period will be the design of network-based monitoring and controlling algorithms, which has already been initiated (Goal 3b). We will also initiate the economic impact assessment (Goal 3c). Goal 4: Uncertainty quantification and sensitivity analysis All the methods developed under the three goals will be rigorously analyzed with regard to robustness. We will analyze the sensitivity of the modeled networks to functional and structural perturbations of the modeled networks. Under these perturbations, we will also analyze how forecasts and outcomes of intervention algorithms change. At least two publications planned for this period: 1. Analysis of food flows using network reliability. 2. Controlling the multi-pathway spread of invasive species using multi-scale network interventions.
Impacts
What was accomplished under these goals?
Goal 1: Data exploration, curation and preliminary analysis In this period, datasets -- vegetable and cereal commodity flows from Freight Analysis Framework (FAF), vegetable production from CROPSCAPE and trade matrix from FAO were downloaded and stored in a PostgreSQL database. Since the data is from multiple sources, we had to standardize the datasets in order to analyze them in combination. Using GIS tools in python programming language, production data was disaggregated to 25kmx25km cells in preparation for use in simulations. Goal 2: Modeling and analysis of commodity flows. Two sets of commodity flow networks were constructed from the above mentioned datasets -- US vegetable and cereal flows from FAF database for various years and country-to-country vegetable flow networks from FAO trade matrix. Structural and dynamical analysis of these datasets are underway. Important nodes in the network such as hubs, sources and sinks were identified based on structural analysis such as indegree, outdegree, betweenness centrality, etc. We are developing a novel approach to identify important dynamics-induced clusters of highly-connected nodes in a directed weighted network using Moore-Shannon network reliability. Goal 3: Multi-pathway models for invasive species spread. The multi-pathway model from our previous work [McNitt et al. 2019] is being extended. Firstly, based on feedback received more functional relationships are being implemented. Secondly, using vectorized operations in python programming language we have been making the simulator faster and therefore scalable to larger networks. This will enable us to run simulations at the region/country scale in the US. We are developing a multi-scale intervention framework with the objective of selecting few locations to setup traps or apply interventions such as pesticides in order to delay or stifle the spread of the pest in the event of its introduction. The approach will apply agent-based models along with linear optimization techniques. McNitt, J., Chungbaek, Y. Y., Mortveit, H., Marathe, M., Campos, M. R., Desneux, N., ... & Adiga, A. (2019). Assessing the multi-pathway threat from an invasive agricultural pest: Tuta absoluta in Asia. Proceedings of the Royal Society B, 286(1913), 20191159.
Publications
|
|