DSpace@RPI (Rensselaer Polytechnic Institute)
Not a member yet
6809 research outputs found
Sort by
Multi-dimensional origin-destination freight flow prediction via a hybrid multi-graph convolutional neural networks based model
No full textDecember 2021School of EngineeringWith the rapid development of eCommerce, freight plays an increasingly important role in contemporary society. The global pandemic further highlights the significance of the freight system. Nowadays, people gradually place more and more reliance and expectations on online shopping, making freight activities more complex. As an important and complicated topic, most existing freight flow prediction studies are limited. For example, the studies either focus on linear temporal dependencies or focus on spatial-temporal dependencies. A multi-dimensional freight flow prediction model is needed to disentangle the freight flow variation pattern better. Additionally, many new freight alternatives emerge thanks to the development of information technology and artificial intelligence. Massive freight data is available, enabling the application of advanced machine learning and deep learning (DL) models. In order to provide a comprehensive and accurate model specialized in freight flow, a multi-graph convolutional based DL model framework is proposed to predict freight flow between origins and destinations considering impacts from the dimensions of temporal, spatial, and socioeconomic features. The DL model consists of three main parts: a long-short term memory module to capture temporal dependencies, a two-dimensional graph convolutional module to capture spatial and socioeconomic features, and a graph fusion module to integrate impacts from multiple dimensions. Following this, an input-output (I-O) table is disaggregated into a lower spatial level according to the distribution of predicted freight flow among different commodities. An empirical application of the proposed model framework is analyzed using a real-world dataset collected from a leading crowdsourcing freight company in China. The case study predicts the crowdsourcing freight flow of each commodity type between each pair of cities in China, including dependencies of time, geographic location, and city characteristics. China’s I-O table is further disaggregated into a city-level I-O table to reveal the inter-commodity dependencies between some city pairs. Unique features of the studied dataset are discussed, and the prediction power of the proposed model is demonstrated. As a result, this dissertation contributes to freight flow studies by offering an end-to-end multi-dimensional DL model framework to predict freight flow OD matrix, and an efficient alternative to disaggregate I-O table to reveal inter-commodity dependencies between smaller spatial units.Ph
Averting a.i. catastrophe : improving democratic intelligence for technological risk governance
No full textAugust 2019School of Humanities, Arts, and Social SciencesConcerns about the negative social impacts of artificial intelligence (AI) continue to grow as rapid technological developments bring the promises and threats of AI into reality. Though long dismissed by AI scientists, developers, and entrepreneurs as irrational fears of an ignorant public duped by an unscrupulous media, public concerns are being borne out as a growing body of evidence suggests that AI, as now practiced, poses significant risks to a majority of humankind. What are the risks of AI, and who is creating them? Reliance on technical experts for the definition of relevant categories carries with it the risk of reproducing both the “hype” surrounding AI and experts’ exclusive focus on technological, rather than sociological, sources of risk. I therefore take a political approach to risk, broadening my focus to include the activities of the creators, owners, and users of AI, as well as those whom they impact. Through participant observation at AI conferences, semi-structured interviews with experts, and textual analysis of primary and secondary literature, my dissertation examines how AI scientists, developers, entrepreneurs, funders, and users create risks, what those risks are, and who they put at risk. I organize this empirical data into seven dimensions of what I call the “AI risk horizon”: military, political, economic, social, environmental, psycho-physiological, and existential risk. Drawing from STS literatures on the governance of technology, I show how risks in all seven dimensions of the horizon emerge from the technocratic political structure of decision making processes in AI research and development. Despite endangering a majority of people, a minority of elites stand to benefit marvelously from AI. In short, one person’s risk is another’s profit My central question is then: What can be done to intervene and mitigate the scope and magnitude of these risks? This dissertation uses a twenty-point framework to evaluate barriers to better risk governance and propose strategies for overcoming them.Ph
Comprehensive, explainable, ml-based molecular toxicity and protein-ligand binding predictions
No full textAugust 2022School of EngineeringMachine Learning (ML) and computer aided drug design have widely been considered to accelerate and focus the time-consuming and costly drug discovery process. In this work, we have used ML tools to improve early-stage drug discovery processes by creating a comprehensive and explainable framework in predicting clinical toxicity of molecules and characterizing structural basis of protein-ligand interactions. Explainable machine learning for molecular toxicity prediction is a promising approach for efficient drug development and chemical safety. A predictive ML model of toxicity can reduce experimental cost and time while mitigating ethical concerns by significantly reducing animal and clinical testing. Herein, we used a deep learning framework for simultaneously modeling in vitro, in vivo, and clinical toxicity data. Two different molecular input representations were used; Morgan fingerprints and pre-trained SMILES embeddings. A multi-task deep learning model accurately predicted toxicity for all endpoints, including clinical, as indicated by the area under the Receiver Operator Characteristic curve and balanced accuracy. In particular, pre-trained molecular SMILES embeddings as input to the multi-task model improved clinical toxicity predictions compared to existing models in MoleculeNet benchmark. Additionally, our multitask approach is comprehensive in the sense that it is comparable to state-of-the-art approaches for specific endpoints in in vitro, in vivo and clinical platforms. Through both the multi-task model and transfer learning, we were able to indicate the minimal need of in vivo data for clinical toxicity predictions. To provide confidence and explain the model’s predictions, we adapted a post-hoc contrastive explanation method that returns pertinent positive and negative features, which correspond well to known mutagenic and reactive toxicophores, such as unsubstituted bonded heteroatoms, aromatic amines, and Michael receptors. Furthermore, toxicophore recovery by pertinent feature analysis captures more of the in vitro (53%) and in vivo (56%), rather than of the clinical (8%), endpoints, and indeed uncovered a preference in known toxicophore data towards in vitro and in vivo experimental data. To our knowledge, this is the first contrastive explanation, using both present and absent substructures, for predictions of clinical and in vivo molecular toxicity. Another nontrivial aspect of the drug discovery process is understanding and characterizing the structural basis of protein-ligand interactions, crucial for developing de novo therapeutics for a protein target. Traditionally, structure-based methods have made tremendous progress over the years, focusing on docking of ligand-protein complexes, adopting classical force-field, empirical or knowledge-based approaches However, these methods rely on the availability of the 3D structure of the given target, and are time consuming and computationally expensive. Machine Learning methods have more recently been applied to predict binding affinity by using existing experimental or computational data. Though these “black-box” models often produce high accuracy predictions they do not provide a human-understandable structural reason for a given prediction. In this work, we created a novel framework to explain the predicted binding of a molecule by a black-box deep neural network model. The framework builds upon the contrastive explanations method (CEM) and provides explanations that go beyond mere correlations identifying minimally sufficient substructures (pertinent positives) that recover the prediction of the black-box, as well as minimal additions (pertinent negatives) that would be necessary to alter its decision. We applied our framework to explain the binding of computationally generated small drug-like small molecules designed by a deep learning model, to three SARS-CoV-2 therapeutic targets, namely the non-structural protein 9 replicase (NSP9), main protease (MPRO) and receptor binding domain of the spike protein (chimeric RBD), which to the best of our knowledge is novel. Pertinent substructures obtained from 1D molecular representations were in agreement with 2D and 3D docking interactions and known pharmacophores. We believe this approach will provide confidence and molecular understanding to high performing ML models predicting protein-ligand interactions, while indicating pertinent substructures to add or avoid in designing ligands. Thus, in this work we have leveraged ML-based approaches to help accelerate different aspects of the drug discovery process. We have provided improved and explainable predictions of clinical toxicity of molecules, and a low computational, explainable, approach to predict structural basis of protein-ligand binding.Ph
Characterization of heparin’s conformational ensemble by molecular dynamics simulations and nuclear magnetic resonance spectroscopy
No full textJournal of Chemical Theory and Computation, 18, 1894-1904Note : if this item contains full text it may be a preprint, author manuscript, or a Gold OA copy that permits redistribution with a license such as CC BY. The final version is available through the publisher’s platform.Heparin is a highly charged, polysulfated polysaccharide and serves as an anticoagulant. Heparin binds to multiple proteins throughout the body, suggesting a large range of potential therapeutic applications. Although its function has been characterized in multiple physiological contexts, heparin’s solution conformational dynamics and structure–function relationships are not fully understood. Molecular dynamics (MD) simulations facilitate the analysis of a molecule’s underlying conformational ensemble, which then provides important information necessary for understanding structure–function relationships. However, for MD simulations to afford meaningful results, they must both provide adequate sampling and accurately represent the energy properties of a molecule. The aim of this study is to compare heparin’s conformational ensemble using two well-developed force fields for carbohydrates, known as GLYCAM06 and CHARMM36, using replica exchange molecular dynamics (REMD) simulations, and to validate these results with NMR experiments. The anticoagulant sequence, an ultra-low-molecular-weight heparin, known as Arixtra (fondaparinux, sodium), was simulated with both parameter sets. The results suggest that GLYCAM06 matches experimental nuclear magnetic resonance three-bond J-coupling values measured for Arixtra better than CHARMM36. In addition, NOESY and ROESY experiments suggest that Arixtra is very flexible in the sub-millisecond time scale and does not adopt a unique structure at 25 C. Moreover, GLYCAM06 affords a much more dynamic conformational ensemble for Arixtra than CHARMM36.National Institutes of Healt
In situ transmission electron microscopy of high-temperature inconel-625 corrosion by molten chloride salts
No full textDecember 2021School of EngineeringHigh-temperature corrosion of molten salt containment materials is of great significance for thermal energy storage systems that are used in concentrated solar power plants (CSP). Mitigating this corrosion is critical for developing cost-effective, energy-efficient systems, which demands comprehensive and thorough understanding of the determinant corrosion reaction mechanisms. So, in this research work, corrosion of Inconel-625 by pure molten chloride salts (MgCl2 − NaCl − KCl) at 500-800 °C has been investigated. It is based on an environmental cell assembly in situ transmission electron microscope (TEM). TEM diffraction and imaging techniques are used to investigate microstructural and compositional evolution at salt-alloy interfaces. A clustering algorithm and a 2D Gaussian fit function are used to determine diffraction spot intensities in in situ diffraction patterns, to quantify alloy corrosion. This facilitates quantitative observation of the evolution of individual grains, in contrast to conventional macroscopic corrosion rate quantification. Procedures are established to minimize incorporation of H2O or O2 from atmosphere in the chloride salts during sample fabrication and corrosion. At first, the corrosive effect of air-exposure on corrosion of Inconel-salt sample is studied by comparing sample corroded with and without air contamination (i.e., vacuum transferred). The Inconel-625 corrosion rate for vacuum transferred samples is 220 ± 20 µm year-1 at 700 °C. Air contamination causes a much more pronounced increase to 1000 ± 150 µm year-1 at 700 °C. Moreover, the individual corrosive effects of the major corroding components present in air that are O2 and H2O vapor are also studied. To perform corrosion in O2 ambient, I employed a top chip that has a gas channel to hold ambient during corrosion. And to study effects of H2O ambient on corrosion, I developed a method to controllably hydrate the salt-stack without exposing it to molecular O2, in a reaction chamber maintained under high vacuum. Then, I investigate corrosion of Inconel-625 by pure molten chloride salts (MgCl2 − NaCl − KCl) at 500-800 °C in 1.0 atm inert N2 or pure O2, or H2O ambient. The isothermal corrosion rates of Inconel-625 are found to be 203 ± 30 μm year-1 for a sample at 700 °C with 1 atm N2 that increased to 463 ± 30 μm year-1 at 800 °C. The corrosion rate increased to 1261 ± 170 μm year-1 for 1 atm O2 ambient at 700 °C. The rate of corrosion in case of a hydrated salt sample at 500 °C is 95 ± 20 μm year-1 that increased to 468 ± 30 μm year-1 at 600 °C and soared to a lower limit of corrosion rate of 3 x 104 μm year-1 at 700 °C. These isothermal corrosion rates indicate that the molten chloride corrosion is significantly accelerated by salt hydration at temperatures above 600 °C. The corrosion is increased at high temperatures due to the generation of increased amounts of corrosive Cl2 and HCl gases coupled with volatile metal compounds. Real time imaging of the microstructure evolution suggests that corrosion is initiated at grain boundaries. Post-corrosion compositional analysis is performed using XPS high resolution scans and AES survey scans and major corrosion products are identified for inert N2 or pure O2, or H2O ambients.Ph
Drug and gene delivery from electrospun fibers for neural repair
No full textMay 2022School of EngineeringPeripheral nerve injury (PNI) affects millions of individuals in the United States alone and can result in loss of motor and sensory function. The native repair response in the peripheral nervous system (PNS) enables peripheral nerve regeneration over short distances. The regenerative capacity of the PNS is often attributed to the immense plasticity of Schwann cells, the primary glia (neuronal support cell) present in the PNS. Following injury, mature Schwann cells can shift into a repair cell phenotype to better support regeneration. In cases of traumatic PNI, Schwann cells may fail to sustain this repair phenotype for the duration necessary and, thus, clinical intervention is required to enable complete regeneration and functional restoration. Synthetic biomaterial nerve grafts are readily investigated to bridge traumatic PNI gaps and improve regenerative outcomes. However, synthetic nerve grafts require further optimization to address the complex pathophysiology of the PNI environment and enable robust regeneration and function restoration. Electrospun fibers are often incorporated into synthetic nerve grafts to improve their regenerative capacity for preclinical studies. The diameter and orientation of electrospun fibers can be modified during the electrospinning process to produce a fibrous mat or graft filler that better mimics the native PNS extracellular matrix (ECM) to support and guide regenerating tissue. However, electrospun fibers alone are often unable to facilitate complete regeneration across critical length injury gaps. Loading the electrospun fibers with therapeutic molecules such as drugs, proteins, and nucleic acids that modulate the Schwann cell response and directly influence axonal regeneration can further improve the regenerative capacity of electrospun fibers. This thesis aims to develop aligned electrospun fiber-mediated drug and gene delivery platforms to improve the regenerative potential of the electrospun fibers for future use in synthetic peripheral nerve grafts.
First, we fabricated and characterized aligned electrospun poly(lactic-co-glycolic acid) (PLGA) fibers encapsulating the immunomodulatory drug fingolimod. Fingolimod holds the potential to target the complex pathophysiology of PNI by stimulating a repair Schwann cell phenotype and endogenous trophic factor production and neurite outgrowth from neurons. In vitro characterization revealed that the fingolimod-releasing electrospun fibers provided sustained release of fingolimod for 28 days, increased neurite outgrowth from whole dorsal root ganglia (DRG) explants and dissociated DRG neurons, increased Schwann cell migration outwards from the DRG body, and decreased Schwann cell mRNA expression levels of several myelin-associated factors. However, the fingolimod-releasing fibers did not increase Schwann cell mRNA expression levels of repair-associated, pro-regenerative factors. These findings indicate that the aligned fingolimod-releasing electrospun fibers likely released enough fingolimod to increase neurite outgrowth and Schwann cell migration but not enough to stimulate a repair Schwann cell phenotype. To our knowledge, this is the first electrospun fiber fingolimod delivery system designed for PNI applications.
Next, we aimed to further enhance the regenerative capacity of the aligned electrospun fibers for PNI applications by creating an aligned electrospun fiber platform for local delivery of mRNA encoding neurotrophin-3 (NT-3). NT-3 holds the potential to target the complex pathophysiology of PNI by supporting Schwann cell and neuron survival, stimulating Schwann cell migration, enhancing neurite outgrowth and axonal regeneration, and supporting remyelination of regenerated axons. Additionally, delivery of mRNA enables the transient expression and secretion of NT-3 protein local to the injury and bypasses several limitations presented by the delivery of proteins that possess low stability and, thus, a short half-life. First, we demonstrated the successful synthesis of bioactive mRNA encoding NT-3 and complexed the mRNA to a cationic transfection agent to form a cationic lipoplex. Next, we immobilized the mRNA lipoplexes to the surface of electrospun fibers with high efficiency through electrostatic interactions, enabling the sustained release of the mRNA lipoplexes over 28 days. In vitro characterization revealed that the aligned electrospun fiber-mediated mRNA delivery platform sustained increased levels of NT-3 protein secretion from Schwann cells for 21 days and stimulated increased neurite outgrowth from whole DRG explants. To our knowledge, this is the first electrospun fiber mRNA delivery system designed for tissue engineering applications. The findings detailed in this thesis provide a basis for further optimization of these electrospun fiber-mediated drug and gene delivery platforms for future testing in a preclinical PNI model.Ph
Spike-encoding of auditory stimuli : modeling studies of pitch and timbre
No full textDecember 2021School of ArchitectureTwo studies are presented that relate psychoacoustic phenomena to neural coding mechanisms. The first describes a new method for estimating pitch from an autocorrelation-like representation of an auditory signal. Specifically, the autocorrelation representation is subjected to a smoothing operation before pitch estimation. This modification enables prediction of the pitch of a stimulus with a mistuned harmonic, a phenomenon that had previously presented a challenge to autocorrelation-based approaches. The resulting model retains the ability to predict many other pitch phenomena. A testable physiological mechanism is proposed that may underlie the pitch shift of a mistuned harmonic complex. This mechanism relies on the dynamics of synchrony-enhancing neurons in the cochlear nucleus, and it is demonstrated that a model of such a neuron, when responding to nearly harmonic stimuli, produces interspike intervals that dilate or contract in a manner that is qualitatively consistent with the pitch shift. The second study concerns the contribution of onset transients to the perception of timbre. It is proposed that information about stimulus onsets is encoded in the precise volley patterns of cortical onset responses. To test this idea, a model of the auditory periphery is fed into a reservoir of spiking neurons that has been designed to capture several important characteristics of a local cortical circuit: realistic spiking dynamics, conduction delays, a rich collection of recurrent connections, and synapses that adapt according to spike-timing-dependent plasticity. Correlational analysis of the model’s responses demonstrates that precise spike patterns corresponding to particular stimulus categories do develop, and spike-timing-dependent plasticity reinforces these patterns. A latency classifier that operates on this information is constructed, and its performance is evaluated against a related study. The spike pattern approach is seen to be more informative in many cases than previously published results. The generalization ability of the classifier is tested, and the results indicate that it is capable of generalizing well on new stimuli provided that onset characteristics are not disturbed. Finally, it is shown that the classifier relies on precise spike patterns rather than overall firing rates.Ph
Corrosion of steel plate girder bridges and rehabilitation using UHPC
No full textAugust 2020School of EngineeringSteel plate girder bridges with concrete slab decks have been widely used in the USA highway network. The average age of all US bridges is around 43 years, including steel plate girder bridges. Today, about 26% of the nation’s bridges are structurally deficient. The concrete slabs suffer from delamination and excessive cracking due to corrosion and acid attacks. The steel girders suffer from sectional losses in different locations due to corrosion. This problem negatively influences on the US economy by imposing large funding for rehabilitation as well as imposing limitations on the transport network efficiency due to the reduction of deficient bridges load rating. Therefore, two main aspects have been considered in this work. The first deals with the accurate prediction of corrosion effects on the steel structure. The second deals with offering middle-ground solutions to extend the bridge life without fully replacing it nor reducing its rated loading. Today, the concrete slab is replaced every 40 years on average, and the girder is strengthened multiple times during regular maintenance, then, the bridge rated load is eventually reduced until it is replaced. The goal of this study was to address both aspects by understanding the steel corrosion more clearly and understanding the behavior of reinforced Ultra-High Performance Concrete to propose it as a novel lighter weight replacement for the slab. On the corrosion side, the effects of different environmental exposure conditions on corrosion morphology and the corresponding reduction in strength and apparent ductility are considered. These conditions included: 1) the presence of chloride ions, 2) Temperature variations, and 3) Mill scale effects. The chloride contents were: 0%, 1%, 2%, and 3% by weight relative to de-ionized water. The containers with the designated chloride contents were stored in rooms with different controlled temperatures of 26C, 40C, and 50C. Unlike the previously published steel corrosion experiments, the manufactured dog-bone samples were exposed to the solutions without removing the mill scale. The accelerated corrosion experimental program lasted for 16 months allowing us to reach corrosion levels corresponding to more than 100 years of real service life. An extended experimental program was performed on coupon samples with different surface conditions: with mill scale (as-received samples) and without mill scale (polished samples) to further investigate the effect of mill scale on corrosion initiation and corroded surface morphology. The extended corrosion experiment was done on two chloride concentrations (0% and 1%) at room temperature (26C) and lasted for three months. All dog-bone samples (corroded and reference) were mechanically tested under uniaxial tension tests. Finite element simulations were performed on the corroded and reference steel dog-bone samples, which show a good agreement with the experimental results and explain the origins of apparent ductility degradation. Among major findings was that the mill scale can initiate corrosion even if there are no chloride ions in the solution. Additionally, corrosion morphology highly affects the ductility, which considers all simplified designs of bridges based on average area losses inaccurate in terms of ductility. On the rehabilitation side, replacement of the RC slab using a much lighter UHPC slab is proposed to reduce the slab weight in proportion to the reduction in girder capacity due to corrosion. A case study on a 43 years old corroded non-composite steel plate girder bridge in the State of Pennsylvania was utilized to prove the feasibility of the proposed method where it was shown that the UHPC can reduce the dead load by more than 20% allowing the bridge to still perform at its full live load demand without the need of rehabilitating the steel structure. Next, a new method for casting UHPC slabs with preferential fiber orientation is developed, as this was shown to be a major contributing parameter to the strength of the slab. Finally, in order to better understand the behavior of UHPC sections and due to the limited design considerations available in literature for reinforced UHPC elements, a comprehensive experimental program was conducted to reveal and understand the interplay between fibers content and reinforcement ratio and how that affects the mode of failure, ultimate strength and ductility. The experimental results were compared with the UHPC available codes, and the results showed how current provisions for design are far from accurate predictions. In general, code predictions are too much conservative in shear and are also conservative in bending. On the ductility side, there is a large need to provide a new definition for under-reinforced versus over-reinforced section design because the interplay of fibers and reinforcement can result in lower overall element ductility if high fiber content is provided with low reinforcement ratios. This study serves as a call for developing performance based measures for sectional designs that include the different reinforcing modalities along with element dimensions and loading conditions.Ph
An Ontology for Fairness Metrics
No full textRecent research has revealed that many machine-learning models and the datasets they are trained on suffer from various forms of bias, and a large number of different fairness metrics have been created to measure this bias. However, determining which metrics to use, as well as interpreting their results, is difficult for a non-expert due to a lack of clear guidance and issues of ambiguity or alternate naming schemes between different research papers. To address this knowledge gap, we present the Fairness Metrics Ontology (FMO), a comprehensive and extensible knowledge resource that defines each fairness metric, describes their use cases, and details the relationships between them. We include additional concepts related to fairness and machine learning models, enabling the representation of specific fairness information within a resource description framework (RDF) knowledge graph. We evaluate the ontology by examining the process of how reasoning-based queries to the ontology were used to guide the fairness metric-based evaluation of a synthetic data model
Performance and functionality objectives in low to mid-rise buildings : a resilience study for wind hazards
No full textMay 2022School of EngineeringWith an increasing frequency and severity of windstorms, design for resilience and achieving functional objectives for the built environment is becoming more critical. The concept of Performance objective assessment is recently extended to Wind Engineering. This approach is applied using the Database-Assisted Design technique, relying on the aerodynamic database provided by the National Institute of Standards and Technology (NIST). The procedure is then repeated for several wind directions and different dominant opening scenarios to determine the cases that meet or exceed performance objective criteria. A framework for recovery-based design in wind engineering is introduced. Currently, research is well advanced to implement this approach in earthquake engineering, with the scope of improving the resilience of structures, critical infrastructures (power plants and grid, water pipelines etc.), to achieve a durable and re-occupiable built environment after the occurrence of a disaster and to provide a tool to estimate the downtimes and quantify the service disruption. The concept of interdependencies between the nonstructural components and the building services is realized using Fault-Tree Analysis (FTA) and applied to four different hospitals with specific focus on the rooms in the zones assumed to be most vulnerable to wind damage. Two actual cases are analyzed that were damaged by hurricanes Maria in 2017, and Michael in 2018. The documentation of post-hazard damage is essential to develop and validate recovery models. In an effort to understand windstorm damage and its impact on recovery, a virtual damage detection and evaluation in post-windstorm reconnaissance is applied to a tornado that touched down in the city of Monroe, LA on April 12, 2020, causing damage to roughly 460 homes in addition to an estimated $250 million of property damage. Multiple reconnaissance teams documented the damage caused by this severe event by capturing video recordings from a drone and imagery from a vehicle-mounted camera. A Streetview application was created using the georeferenced panoramic images. Multiple evaluators reviewed datasets collected separately to determine a damage state and associated estimated wind speed using the current EF-Scale (WISE, 2006) and draft language from the EF-Scale chapter in the forthcoming ASCE/SEI/AMS standard on wind speed estimation. The large database of damage assessments then allowed comparisons between different evaluators, different modalities (ground-based and remote-sensing), and between the current and proposed revised EF-Scale. Particular attention is given to the limitations of each reconnaissance technology and the implications of each of the EF-Scales.Ph