-
Stacked Autoencoder Based Feature Extraction and Superpixel Generation for Multifrequency PolSAR Image Classification
Authors:
Tushar Gadhiya,
Sumanth Tangirala,
Anil K. Roy
Abstract:
In this paper we are proposing classification algorithm for multifrequency Polarimetric Synthetic Aperture Radar (PolSAR) image. Using PolSAR decomposition algorithms 33 features are extracted from each frequency band of the given image. Then, a two-layer autoencoder is used to reduce the dimensionality of input feature vector while retaining useful features of the input. This reduced dimensional…
▽ More
In this paper we are proposing classification algorithm for multifrequency Polarimetric Synthetic Aperture Radar (PolSAR) image. Using PolSAR decomposition algorithms 33 features are extracted from each frequency band of the given image. Then, a two-layer autoencoder is used to reduce the dimensionality of input feature vector while retaining useful features of the input. This reduced dimensional feature vector is then applied to generate superpixels using simple linear iterative clustering (SLIC) algorithm. Next, a robust feature representation is constructed using both pixel as well as superpixel information. Finally, softmax classifier is used to perform classification task. The advantage of using superpixels is that it preserves spatial information between neighbouring PolSAR pixels and therefore minimises the effect of speckle noise during classification. Experiments have been conducted on Flevoland dataset and the proposed method was found to be superior to other methods available in the literature.
△ Less
Submitted 6 November, 2023;
originally announced November 2023.
-
${\tt MORALS}$: Analysis of High-Dimensional Robot Controllers via Topological Tools in a Latent Space
Authors:
Ewerton R. Vieira,
Aravind Sivaramakrishnan,
Sumanth Tangirala,
Edgar Granados,
Konstantin Mischaikow,
Kostas E. Bekris
Abstract:
Estimating the region of attraction (${\tt RoA}$) for a robot controller is essential for safe application and controller composition. Many existing methods require a closed-form expression that limit applicability to data-driven controllers. Methods that operate only over trajectory rollouts tend to be data-hungry. In prior work, we have demonstrated that topological tools based on…
▽ More
Estimating the region of attraction (${\tt RoA}$) for a robot controller is essential for safe application and controller composition. Many existing methods require a closed-form expression that limit applicability to data-driven controllers. Methods that operate only over trajectory rollouts tend to be data-hungry. In prior work, we have demonstrated that topological tools based on ${\it Morse Graphs}$ (directed acyclic graphs that combinatorially represent the underlying nonlinear dynamics) offer data-efficient ${\tt RoA}$ estimation without needing an analytical model. They struggle, however, with high-dimensional systems as they operate over a state-space discretization. This paper presents ${\it Mo}$rse Graph-aided discovery of ${\it R}$egions of ${\it A}$ttraction in a learned ${\it L}$atent ${\it S}$pace (${\tt MORALS}$). The approach combines auto-encoding neural networks with Morse Graphs. ${\tt MORALS}$ shows promising predictive capabilities in estimating attractors and their ${\tt RoA}$s for data-driven controllers operating over high-dimensional systems, including a 67-dim humanoid robot and a 96-dim 3-fingered manipulator. It first projects the dynamics of the controlled system into a learned latent space. Then, it constructs a reduced form of Morse Graphs representing the bistability of the underlying dynamics, i.e., detecting when the controller results in a desired versus an undesired behavior. The evaluation on high-dimensional robotic datasets indicates data efficiency in ${\tt RoA}$ estimation.
△ Less
Submitted 15 March, 2024; v1 submitted 4 October, 2023;
originally announced October 2023.
-
Roadmaps with Gaps over Controllers: Achieving Efficiency in Planning under Dynamics
Authors:
Aravind Sivaramakrishnan,
Sumanth Tangirala,
Edgar Granados,
Noah R. Carver,
Kostas E. Bekris
Abstract:
This paper aims to improve the computational efficiency of motion planning for mobile robots with non-trivial dynamics through the use of learned controllers. It adopts a decoupled strategy, where a system-specific controller is first trained offline in an empty environment to deal with the robot's dynamics. For a target environment, the proposed approach constructs offline a data structure, a "Ro…
▽ More
This paper aims to improve the computational efficiency of motion planning for mobile robots with non-trivial dynamics through the use of learned controllers. It adopts a decoupled strategy, where a system-specific controller is first trained offline in an empty environment to deal with the robot's dynamics. For a target environment, the proposed approach constructs offline a data structure, a "Roadmap with Gaps," to approximately learn how to solve planning queries in this environment using the learned controller. The nodes of the roadmap correspond to local regions. Edges correspond to applications of the learned control policy that approximately connect these regions. Gaps arise because the controller does not perfectly connect pairs of individual states along edges. Online, given a query, a tree sampling-based motion planner uses the roadmap so that the tree's expansion is informed towards the goal region. The tree expansion selects local subgoals given a wavefront on the roadmap that guides towards the goal. When the controller cannot reach a subgoal region, the planner resorts to random exploration to maintain probabilistic completeness and asymptotic optimality. The accompanying experimental evaluation shows that the approach significantly improves the computational efficiency of motion planning on various benchmarks, including physics-based vehicular models on uneven and varying friction terrains as well as a quadrotor under air pressure effects.
△ Less
Submitted 28 March, 2024; v1 submitted 4 October, 2023;
originally announced October 2023.
-
Direct numerical simulation of thermo-hydro-mechanical processes at Soultz-sous-Forêts
Authors:
Saeed Mahmoodpour,
Mrityunjay Singh,
Ramin Mahyapour,
Sri Kalyan Tangirala,
Kristian Bär,
Ingo Sass
Abstract:
Porosity and permeability alteration due to the thermo-poro-elastic stress field disturbance from the cold fluid injection is a deciding factor for longer, economic and safer heat extraction from an enhanced geothermal system (EGS). In the Soultz-sous-Forêts geothermal system, faulted zones are the main flow paths and the resulting porosity-permeability development over time due to stress reorient…
▽ More
Porosity and permeability alteration due to the thermo-poro-elastic stress field disturbance from the cold fluid injection is a deciding factor for longer, economic and safer heat extraction from an enhanced geothermal system (EGS). In the Soultz-sous-Forêts geothermal system, faulted zones are the main flow paths and the resulting porosity-permeability development over time due to stress reorientation is more sensitive in comparison with the regions without faulted zones. Available operational and field data are combined through a validated numerical simulation model to examine the mechanical impact on the pressure and temperature evolution. Results shows that near the injection wellbore zones, permeability and porosity values are strongly affected by the stress field changes and the permeability changes will affect the overall temperature and pressure of the system demonstrating a fully coupled phenomenon. In some regions inside the faulted zones and close to the injection wellbores, the porosity doubles whereas permeability may enhance up to 30 times. From the sensitivity analysis on the two unknown parameters from the mechanical aspect is performed and results shows that only one of the parameters impacts significantly on the porosity-permeability changes. Further experimental and field works on this parameter will help to model the heat extraction from the Soultz-sous-Forêts and elsewhere more precisely than before.
△ Less
Submitted 3 June, 2022;
originally announced June 2022.
-
Identifying Critical LMS Features for Predicting At-risk Students
Authors:
Ying Guo,
Cengiz Gunay,
Sairam Tangirala,
David Kerven,
Wei **,
Jamye Curry Savage,
Seung** Lee
Abstract:
Learning management systems (LMSs) have become essential in higher education and play an important role in hel** educational institutions to promote student success. Traditionally, LMSs have been used by postsecondary institutions in administration, reporting, and delivery of educational content. In this paper, we present an additional use of LMS by using its data logs to perform data-analytics…
▽ More
Learning management systems (LMSs) have become essential in higher education and play an important role in hel** educational institutions to promote student success. Traditionally, LMSs have been used by postsecondary institutions in administration, reporting, and delivery of educational content. In this paper, we present an additional use of LMS by using its data logs to perform data-analytics and identify academically at-risk students. The data-driven insights would allow educational institutions and educators to develop and implement pedagogical interventions targeting academically at-risk students. We used anonymized data logs created by Brightspace LMS during fall 2019, spring 2020, and fall 2020 semesters at our college. Supervised machine learning algorithms were used to predict the final course performance of students, and several algorithms were found to perform well with accuracy above 90%. SHAP value method was used to assess the relative importance of features used in the predictive models. Unsupervised learning was also used to group students into different clusters based on the similarities in their interaction/involvement with LMS. In both of supervised and unsupervised learning, we identified two most-important features (Number_Of_Assignment_Submissions and Content_Completed). More importantly, our study lays a foundation and provides a framework for develo** a real-time data analytics metric that may be incorporated into a LMS.
△ Less
Submitted 27 April, 2022;
originally announced April 2022.
-
Identifying the differences between 3 dimensional shapes Using a Custom-built Smart Glove
Authors:
Davis Le,
Sairam Tangirala,
Tae Song Lee
Abstract:
Sensor embedded glove systems have been reported to require careful, time consuming and precise calibrations on a per user basis in order to obtain consistent usable data. We have developed a low cost, flex sensor based smart glove system that may be resilient to the common limitations of data gloves. This system utilizes an Arduino based micro controller as well as a single flex sensor on each fi…
▽ More
Sensor embedded glove systems have been reported to require careful, time consuming and precise calibrations on a per user basis in order to obtain consistent usable data. We have developed a low cost, flex sensor based smart glove system that may be resilient to the common limitations of data gloves. This system utilizes an Arduino based micro controller as well as a single flex sensor on each finger. Feedback from the Arduinos analog to digital converter can be used to infer objects dimensional properties, the reactions of each individual finger will differ with respect to the size and shape of a grasped object. In this work, we report its use in statistically differentiating stationary objects of spherical and cylindrical shapes of varying radii regardless of the variations introduced by gloves users. Using our sensor embedded glove system, we explored the practicability of object classification based on the tactile sensor responses from each finger of the smart glove. An estimated standard error of the mean was calculated from each of the of five fingers averaged flex sensor readings. Consistent with the literature, we found that there is a systematic dependence between an objects shape, dimension and the flex sensor readings. The sensor output from at least one finger, indicated a non-overlap** confidence interval when comparing spherical and cylindrical objects of the same radius. When sensing spheres and cylinders of varying sizes, all five fingers had a categorically varying reaction to each shape. We believe that our findings could be used in machine learning models for real-time object identification.
△ Less
Submitted 8 January, 2022;
originally announced January 2022.
-
The Dynamic Scaling Study of Vapor Deposition Polymerization: A Monte Carlo Approach
Authors:
Sairam Tangirala,
David P. Landau,
Y. -P. Zhao
Abstract:
The morphological scaling properties of linear polymer films grown by vapor deposition polymerization (VDP) are studied by 1+1D Monte Carlo simulations. The model implements the basic processes of random angle ballistic deposition ($F$), free-monomer diffusion ($D$) and monomer adsorption along with the dynamical processes of polymer chain initiation, extension, and merger. The ratio $G=D/F$ is fo…
▽ More
The morphological scaling properties of linear polymer films grown by vapor deposition polymerization (VDP) are studied by 1+1D Monte Carlo simulations. The model implements the basic processes of random angle ballistic deposition ($F$), free-monomer diffusion ($D$) and monomer adsorption along with the dynamical processes of polymer chain initiation, extension, and merger. The ratio $G=D/F$ is found to have a strong influence on the polymer film morphology. Spatial and temporal behavior of kinetic roughening has been extensively studied using finite-length scaling and height-height correlations $H(r,t)$. The scaling analysis has been performed within the no-overhang approximation and the scaling behaviors at local and global length scales were found to be very different. The global and local scaling exponents for morphological evolution have been evaluated for varying free-monomer diffusion by growing the films at $G$ = $10$, $10^2$, $10^3$, and $10^4$ and fixing the deposition flux $F$. With an increase in $G$ from $10$ to $10^4$, the average growth exponent $β\approx 0.50$ was found to be invariant, where as the global roughness exponent $α_{g}$ decreased from $0.87(1)$ to $0.73(1)$ along with a corresponding decrease in the global dynamic exponent $z_g$ from $1.71(1)$ to $1.38(2)$. The global scaling exponents were observed to follow the dynamic scaling hypothesis, $z_g=α_{g}/β$. With a similar increase in $G$ however, the average local roughness exponent ${α_{l}}$ remained close to $0.46$ and the anomalous growth exponent $β_{\ast}$ decreased from 0.23(4) to 0.18(8). The interfaces display anomalous scaling and multiscaling in the relevant height-height correlations. The variation of $H(r,t)$ with deposition time $t$ indicates non-stationary growth.
△ Less
Submitted 22 October, 2010;
originally announced October 2010.
