The application of feedback to enhance motivation is beneficial across various life contexts. While both feedback and motivation have been studied widely in psychological science, most of this research has used close-ended approaches to study feedback empirically, which limits the scope of investigation. The present study was one of the first applications of text-analysis to assess the impact of feedback on the recipient's motivation. A transformer machine-learning model was used to create a tool that can predict the average motivating influence of a particular feedback statement, as perceived by a recipient within an academic context. Feedback was defined and evaluated from the perspective of Feedback Intervention Theory (FIT). Both research hypotheses were supported, given that the model's motivation predictions were positively associated with the actual motivation scores of feedback statements, and the model was closer to estimating the true motivation scores than expected by chance. These findings, paired with additional exploratory analyses, demonstrated the utility and effectiveness of the model in predicting perceived student motivation from feedback statements. Thus, this research provided a reliable tool researchers and practitioners in academia could use to evaluate the motivating influence of feedback for students, and it might inspire future studies in this domain. / Doctor of Philosophy / The use of feedback to enhance motivation is beneficial across various life domains. While both feedback and motivation have been studied widely in psychological science, most of this research has used close-ended (not text-analytic) approaches to study feedback empirically, which limits the scope of investigation. The present study was one of the first applications of text-analysis to assess the impact of feedback on the recipient's motivation. A machine-learning model was used to create a tool that can predict the average motivating influence of a particular feedback statement, as perceived by a recipient within an academic context. Both research hypotheses were supported. The motivation predictions were positively associated with the actual motivation scores of feedback statements, and the model was closer to estimating the true motivation scores than would be expected by chance. These findings, paired with additional exploratory analyses, demonstrated the utility and effectiveness of the model in predicting perceived student motivation from feedback statements. Additionally, based on this study it is recommended that professors include specific behaviors to be modified when delivering feedback. Thus, this research provided a tool that researchers and practitioners in academia could use to evaluate the motivating influence of feedback for students, and it might certainly inspire future studies in this domain.

Bats are famous for their capability of navigating in dense forests for hundreds of kilometers within one night by using their sonar system. Airborne sonar hasn't been heavily used in the industrial world compared to other sensors such as lidar, radar, and cameras. In this study, we applied a biosonar robot to navigate in a dense forest with bat-like FM-CF ultrasonic signals with deep learning. The results presented show that airborne biosonar can classify different areas' plants, in addition to achieving a similar level of navigation granularity compared to GPS, which is about 6 meters of radius resolution. The time- frequency representations of echoes from the forest are used as input data to explore the biosonar navigation ability, and the state-of-the-art CNN deep network (Resnet 152) is used as the brain to do the echolocation in the dense forest. The navigation ability can be improved significantly by combining multiple 10 ms long echoes, however, the data size of the reflected waves is much smaller than the other popularly used sensors, as echo can be collected at a rate of 40 echoes per second. The results can prove that airborne sonar can be used to navigate in GPS-denied environments, and can be an important sensor used in a scenario when other sensors meet constraints, like in the sensor fusion applications. / Doctor of Philosophy / The ability to identify natural landmarks could contribute to the navigation skills of echolo- cating bats and also advance the quest for autonomy in natural environments with man- made systems. The critical sensors used in autonomous robot navigation are camera array, radar, and lidar, airborne sonar hasn't been verified for its navigation efficiency. However, recognizing natural landmarks based on biosonar echoes has to deal with the unpredictable nature of echoes that are typically superpositions of contributions from many different reflec- tors with unknown properties. This dissertation intends to explore the bioinspired airborne sonar navigation ability in dense natural forests. The first part of this project is to use reflected echoes to navigate on a large scale, data was collected from different mountains which are dozens of kilometers away from each other, and we achieved the use of one single navigator in those locations. The second part is to explore the navigation granularity of airborne sonar sensors, data were collected from a small dense forest area, we try to classify which part of the foliage was based on the echo, and in the end, we achieved GPS accuracy for navigation. The finding in this work proves that the sonar sensor can play an important role in the sensing system, with the help of a deep neural network, with a 10 ms long echo, it can have a similar navigation ability to GPS.

In this work, we present different deep learning-based digital pre-distorters and compare them based on their performance towards improving the linearity of highly non-linear power amplifiers. The simulation results show that BiLSTM based DPDs work the best in terms of improving the linearity performance. We also compare two methodologies of direct learning and indirect learning to develop deep learning-based digital pre-distorters (DL-DPDs) models and evaluate their improvement on the linearity of Power Amplifiers (PA). We carry out a theoretical analysis on the differences between these training methodologies and verify their performance with simulation results on class-AB and class-F⁻¹ PAs. The simulation results show that both the learning methods lead to an improvement of more than 12 dB and 11dB in the linearity of class-AB and class-F⁻¹ PAs respectively, with indirect learning DL-DPD offering marginally better performance. Moreover, we compare the DL-DPD with memory polynomial models and show that using the former gives a significant improvement over the memory polynomials. Furthermore, we discuss the advantages of exploiting a BiLSTM based neural network architecture for designing direct/indirect DPDs. We demonstrate that BiLSTM DPD can be used to pre distort signals of any size without the drop in linearity. Moreover, based on the insights we develop a frequency domain loss using which further increased the linearity of the PA. / Master of Science / Wireless communication devices have fundamentally changed the way we interact with people. This increased the user's reliance on communication devices and significantly grew the need for higher data rates and faster internet speeds. But one major obstacle inside the transmitter chain (antenna) with increasing the data rates is the power amplifier, which distorts the signals at these higher powers. This distortion will reduce the efficiency and reliability of communication systems, greatly decreasing the quality of communication. So, we developed a high-performance DPD using deep learning to combat this issue. In this paper, we compare different deep learning-based DPDs and analyze which offers better performance. We also contrast two training methodologies to learn these DL-DPDs, theoretically and with simulation to arrive at which method offers better performing DPDs. We do these experiments on two different types of power amplifiers, and signals of any length. We design a new loss function, such that optimizing it leads to better DL-DPDs.

In recent years previously restricted radio-frequency spectrum has been opened to civilian and industrial access in the United States. Because of this, high priority users such as the military and government need to develop systems that can adapt to the surrounding spectral environment which will suddenly be filled with new users. This thesis considers an environment with one tracking radar, a single target, and a communications system that can passively interfere with the radar system. Three separate agents, Sense and Avoid, Machine Learning, and "Optimal", are tasked with the channel selection problem for radar communications coexistence. Each agent is evaluated based on their ability to detect and avoid the interferer while also tracking a target accurately. In particular, in this thesis, we are interested in the value that machine learning algorithms can provide over and above simple approaches. This value is assessed based on the conflicting requirements of avoiding interference yet using as much of the spectrum for tracking as possible. / Master of Science / With a newfound dependence on wireless transmission, the demand for electromagnetic spectrum allocations has vastly increased. In recent years the Federal Communications Commission has auctioned some previously restricted access frequency bands to public and commercial applications. While this enables the growth of faster and more widespread civilian communications, military radar systems which had been the priority users of those bands are now at risk of interference from new users. Current radar systems typically occupy fixed bands and are not yet well adjusted to sharing their allocated spectrum with other users. Cognitive radar systems have been proposed to monitor airwaves for potential interferences and autonomously manage band allocation to avoid the interferers. In this thesis, we study a learning algorithm that enables a radar system to actively monitor and select its bandwidth to ensure proper target tracking. In particular, we are interested in the value this learning algorithm can provide over and above simple approaches. This value is assessed based on the conflicting requirements of avoiding interference yet using as much of the spectrum for tracking as possible.

Horseshoe bats (family Rhinolophidae) are among the bat species that dynamically deform their reception baffles (pinnae) and emission baffles (noseleaves) during signal reception and emissions, respectively. These dynamics are a focus of prior studies that demonstrated that these effects could introduce time-variance within emitted and received signals. Recent lab based experiments with biomimetic hardware have shown that these dynamics can also inject time-variant signatures into echoes from simple targets. However, complex foliage echoes, which comprise a large portion of the received echoes and contain useful information for these bats, have not been studied in prior research. We used a biomimetic sonarhead which replicated these dynamics, to collect a large dataset of foliage echoes (>55,000). To generate a neuromorphic representation of echoes that was representative of the neural spikes in bat brains, we developed an auditory processing model based on Horseshoe bat physiological data. Then, machine learning classifiers were employed to classify these spike representations of echoes into distinct groups, based on the presence or absence of dynamics' effects. Our results showed that classification with up to 80% accuracy was possible, indicating the presence of these effects in foliage echoes, and their persistence through the auditory processing. These results suggest that these dynamics' effects might be present in bat brains, and therefore have the potential to inform behavioral decisions. Our results also indicated that potential benefits from these effects might be location specific, as our classifier was more effective in classifying echoes from the same physical location, compared to a dataset with significant variation in recording locations. This result suggested that advantages of these effects may be limited to the context of particular surroundings if the bat brain similarly fails to generalize over variation in locations. / Master of Science / Horseshoe bats (family Rhinolophidae) are an echolocating bat species, i.e., they emit sound waves and use the corresponding echoes received from the environment to gather information for navigation. This species of bats demonstrate the behavior of deforming their emitter (noseleaf), and ears (pinna), while emitting or receiving echolocation signals. Horseshoe bats are adept at navigating in the dark through dense foliage. Their impressive navigational abilities are of interest to researchers, as their biology can inspire solutions for autonomous drone navigation in foliage and underwater. Prior research, through numerical studies and experimental reproductions, has found that these deformations can introduce time-dependent changes in the emitted and received signals. Furthermore, recent research using a biomimetic robot has found that echoes received from simple shapes, such as cube and sphere, also contain time-dependent changes. However, prior studies have not used foliage echoes in their analysis, which are more complex, since they include a large number of randomly distributed targets (leaves). Foliage echoes also constitute a large share of echoes from the bats' habitats, hence an understanding of the effects of the dynamic deformations on these foliage echoes is of interest. Since echolocation signals exist within bat brains as neural spikes, it is also important to understand if these dynamic effects can be identified within such signal representations, as that would indicate that these effects are available to the bats' brains. In this study, a biomimetic robot that mimicked the dynamic pinna and noseleaf deformation was used to collect a large dataset (>55,000) of echoes from foliage. A signal processing model that mimicked the auditory processing of these bats and generated simulated spike responses was also developed. Supervised machine learning was used to classify these simulated spike responses into two groups based on the presence or absence of these dynamics' effects. The success of the machine learning classifiers of up to 80% accuracy suggested that the dynamic effects exist within foliage echoes and also spike-based representations. The machine learning classifier was more accurate when classifying echoes from a small confined area, as compared to echoes distributed over a larger area with varying foliage. This result suggests that any potential benefits from these effects might be location-specific if the bat brain similarly fails to generalize over the variation in echoes from different locations.

Metagenomics has become a reliable tool for the analysis of the microbial diversity and the molecular mechanisms carried out by microbial communities. By the use of next generation sequencing, metagenomic studies can generate millions of short sequencing reads that are processed by computational tools. However, with the rapid adoption of metagenomics a large amount of data has been generated. This situation requires the development of computational tools and pipelines to manage the data scalability, accessibility, and performance. In this thesis, several strategies varying from command line, web-based platforms to machine learning have been developed to address these computational challenges. Interpretation of specific information from metagenomic data is especially a challenge for environmental samples as current annotation systems only offer broad classification of microbial diversity and function. Therefore, I developed MetaStorm, a public web-service that facilitates customization of computational analysis for metagenomic data. The identification of antibiotic resistance genes (ARGs) from metagenomic data is carried out by searches against curated databases producing a high rate of false negatives. Thus, I developed DeepARG, a deep learning approach that uses the distribution of sequence alignments to predict over 30 antibiotic resistance categories with a high accuracy. Curation of ARGs is a labor intensive process where errors can be easily propagated. Thus, I developed ARGminer, a web platform dedicated to the annotation and inspection of ARGs by using crowdsourcing. Effective environmental monitoring tools should ideally capture not only ARGs, but also mobile genetic elements and indicators of co-selective forces, such as metal resistance genes. Here, I introduce NanoARG, an online computational resource that takes advantage of the long reads produced by nanopore sequencing technology to provide insights into mobility, co-selection, and pathogenicity. Sequence alignment has been one of the preferred methods for analyzing metagenomic data. However, it is slow and requires high computing resources. Therefore, I developed MetaMLP, a machine learning approach that uses a novel representation of protein sequences to perform classifications over protein functions. The method is accurate, is able to identify a larger number of hits compared to sequence alignments, and is >50 times faster than sequence alignment techniques. / Doctor of Philosophy / Antimicrobial resistance (AMR) is one of the biggest threats to human public health. It has been estimated that the number of deaths caused by AMR will surpass the ones caused by cancer on 2050. The seriousness of these projections requires urgent actions to understand and control the spread of AMR. In the last few years, metagenomics has stand out as a reliable tool for the analysis of the microbial diversity and the AMR. By the use of next generation sequencing, metagenomic studies can generate millions of short sequencing reads that are processed by computational tools. However, with the rapid adoption of metagenomics, a large amount of data has been generated. This situation requires the development of computational tools and pipelines to manage the data scalability, accessibility, and performance. In this thesis, several strategies varying from command line, web-based platforms to machine learning have been developed to address these computational challenges. In particular, by the development of computational pipelines to process metagenomics data in the cloud and distributed systems, the development of machine learning and deep learning tools to ease the computational cost of detecting antibiotic resistance genes in metagenomic data, and the integration of crowdsourcing as a way to curate and validate antibiotic resistance genes.

Even under the best management strategies contemporary crops face yield losses from diverse threats such as, pathogens, pests, and environmental stress. Adding to this management challenge is that under current global climate projections these impacts are predicted to become even greater. Natural genetic variation, long used by traditional plant breeders, holds great promise for adapting high performing agronomic lines to these stressors. Yet, efforts to bolster crop plant resilience using wild relatives have been hindered by time consuming efforts to develop genomic tools and/or identify the genetic basis for agronomic traits. Thus, increasing crop plant resilience requires developing and deploying approaches that leverage current high-throughput sequencing technologies to more rapidly and robustly develop genomic tools in these systems. Here we report the integration of bioinformatic and statistical tools to leverage high-throughput sequencing to 1) develop a machine learning approach to determine factors impacting transcriptome assembly and quantitatively evaluate transcriptome completeness, 2) dissect complex physiological pathway interactions in Solanum pimpinellifolium under combined stresses—using comparative transcriptomics, and 3) develop a genome assembly pipeline that can be deployed to rapidly assemble a more contiguous genome, unraveling previously hidden complexity, using Phytopthora capsici as a model. As a result, we have generated strategic guidelines for transcriptome assembly and developed an orthologue and reference free, machine learning based tool "WWMT" to quantitatively score transcriptome completeness from short read data. Secondly, we identified "hub genes" and describe genes involved with "cross-talk" between drought and herbivore stress response pathways. Finally, we demonstrate a protocol for combining long-read sequencing from the Oxford Nanopore Technologies MinION, and short-read data, to rapidly assembly a cost-effective, contiguous and relatively complete genome. Here we uncovered hidden variation in a well-known plant pathogen finding that the genome was 92% bigger than previous estimates with more than 39% of duplicated regions, supporting a hypothesized recent whole genome duplication in this clade. This community resource will support new functional and evolutionary studies in this economically important pathogen. / Doctor of Philosophy / Meeting the food production demands of a burgeoning population in a changing environment, means adapting crop plants to become more resilient to environmental stress. One of the greatest barriers to understanding and predicting crop responses to future environmental change is our poor understanding of the functional and genomic basis of stress resistance traits for contemporary crops. This impediment presents a barrier for rapid crop improvement technologies, such as, gene editing or genomic selection, that is only partially overcome by generating large amounts of sequencing data. Here we need tools that allow us to process and evaluate huge amounts of data generated from next generation sequencing studies to help identify genomic regions associated with agronomic traits. We also need technical approaches that allow us to disentangle the complex genetic interactions that drive plant stress responses. Here we present work that used statistical analysis and recent advances of artificial intelligence to develop a bioinformatic approach to evaluate genomic sequencing data prior to downstream analyses. Secondly, we used a reductionist approach to filter thousands of genes to key genes associated with combined stress responses (herbivory and drought), in the most widely used vegetable in the world, tomato. Finally, we developed a method for generating whole genome sequences that is low-cost and time sensitive and tested it using a well-known plant pathogen genome, wherein we unraveled significant hidden complexity. Overall this work provides community-wide genomic tools and information to promote crop resilience.

In the thesis, we aimed at building a robust UAV(drone) detection algorithm through which, one drone could detect another drone in flight. Though this was a straight forward object detection problem, the biggest challenge we faced for drone detection is the limited amount of drone images for training. To address this issue, we used Generative Adversarial Networks, CycleGAN to be precise, for the generation of realistic looking fake images which were indistinguishable from real data. CycleGAN is a classic example of Image to Image Translation technique, and we this applied in our situation where synthetic images from one domain were transformed into another domain, containing real data. The model, once trained, was capable of generating realistic looking images from synthetic data without the presence of real images. Following this, we employed a state of the art object detection model, YOLO(You Only Look Once), to build a Drone Detection model that was trained on the generated images. Finally, the performance of this model was compared against different datasets in order to evaluate its performance. / Master of Science / In the recent years, technologies like Deep Learning and Machine Learning have seen many rapid developments. Among the many applications they have, object detection is one of the widely used application and well established problems. In our thesis, we deal with a scenario where we have a swarm of drones and our aim is for one drone to recognize another drone in its field of vision. As there was no drone image dataset readily available, we explored different ways of generating realistic data to address this issue. Finally, we proposed a solution to generate realistic images using Deep Learning techniques and trained an object detection model on it where we evaluated how well it has performed against other models.

Email correspondence has become the predominant method of communication for businesses. If not for the inherent privacy concerns, this electronically searchable data could be used to better understand how employees interact. After the Enron dataset was made available, researchers were able to provide great insight into employee behaviors based on the available data despite the many challenges with that dataset. The work in this thesis demonstrates a suite of methods to an appropriately anonymized academic email dataset created from volunteers' email metadata. This new dataset, from an internal email server, is first used to validate feature extraction and machine learning algorithms in order to generate insight into the interactions within the center. Based solely on email metadata, a random forest approach models behavior patterns and predicts employee job titles with $96%$ accuracy. This result represents classifier performance not only on participants in the study but also on other members of the center who were connected to participants through email. Furthermore, the data revealed relationships not present in the center's formal operating structure. The culmination of this work is an organic organizational chart, which contains a fuller understanding of the center's internal structure than can be found in the official organizational chart. / Master of Science

Finding a fire inside of a structure without knowing its conditions poses a dangerous threat to the safety of firefighters. As a result, robots are being explored to increase awareness of the conditions inside structures before having firefighter enter. This thesis presents a method that autonomously guides a robot to the location of a fire inside a structure. The method uses classification of fire, smoke, and other fire environment objects to calculate a weighted probability. Weighted probability is a measurement that indicates the probability that a given region on an infra-red image will lead to fire. This method was tested on large-scale fire videos with a robot moving towards a fire and it is also compared to following the highest temperatures on the image. Sending a robot to find a fire has the potential to save the lives of firefighters. / Master of Science