Matching Items (4)
Filtering by
- All Subjects: Convolutional Neural Networks
- Genre: Academic theses
- Creators: Li, Baoxin
- Creators: Muhich, Christopher
- Resource Type: Text
Description
In this thesis, a new approach to learning-based planning is presented where critical regions of an environment with low probability measure are learned from a given set of motion plans. Critical regions are learned using convolutional neural networks (CNN) to improve sampling processes for motion planning (MP).
In addition to an identification network, a new sampling-based motion planner, Learn and Link, is introduced. This planner leverages critical regions to overcome the limitations of uniform sampling while still maintaining guarantees of correctness inherent to sampling-based algorithms. Learn and Link is evaluated against planners from the Open Motion Planning Library (OMPL) on an extensive suite of challenging navigation planning problems. This work shows that critical areas of an environment are learnable, and can be used by Learn and Link to solve MP problems with far less planning time than existing sampling-based planners.
In addition to an identification network, a new sampling-based motion planner, Learn and Link, is introduced. This planner leverages critical regions to overcome the limitations of uniform sampling while still maintaining guarantees of correctness inherent to sampling-based algorithms. Learn and Link is evaluated against planners from the Open Motion Planning Library (OMPL) on an extensive suite of challenging navigation planning problems. This work shows that critical areas of an environment are learnable, and can be used by Learn and Link to solve MP problems with far less planning time than existing sampling-based planners.
ContributorsMolina, Daniel, M.S (Author) / Srivastava, Siddharth (Thesis advisor) / Li, Baoxin (Committee member) / Zhang, Yu (Committee member) / Arizona State University (Publisher)
Created2019
Description
Cardiovascular disease (CVD) is the leading cause of mortality yet largely preventable, but the key to prevention is to identify at-risk individuals before adverse events. For predicting individual CVD risk, carotid intima-media thickness (CIMT), a noninvasive ultrasound method, has proven to be valuable, offering several advantages over CT coronary artery calcium score. However, each CIMT examination includes several ultrasound videos, and interpreting each of these CIMT videos involves three operations: (1) select three enddiastolic ultrasound frames (EUF) in the video, (2) localize a region of interest (ROI) in each selected frame, and (3) trace the lumen-intima interface and the media-adventitia interface in each ROI to measure CIMT. These operations are tedious, laborious, and time consuming, a serious limitation that hinders the widespread utilization of CIMT in clinical practice. To overcome this limitation, this paper presents a new system to automate CIMT video interpretation. Our extensive experiments demonstrate that the suggested system significantly outperforms the state-of-the-art methods. The superior performance is attributable to our unified framework based on convolutional neural networks (CNNs) coupled with our informative image representation and effective post-processing of the CNN outputs, which are uniquely designed for each of the above three operations.
ContributorsShin, Jaeyul (Author) / Liang, Jianming (Thesis advisor) / Maciejewski, Ross (Committee member) / Li, Baoxin (Committee member) / Arizona State University (Publisher)
Created2016
Description
Computational visual aesthetics has recently become an active research area. Existing state-of-art methods formulate this as a binary classification task where a given image is predicted to be beautiful or not. In many applications such as image retrieval and enhancement, it is more important to rank images based on their aesthetic quality instead of binary-categorizing them. Furthermore, in such applications, it may be possible that all images belong to the same category. Hence determining the aesthetic ranking of the images is more appropriate. To this end, a novel problem of ranking images with respect to their aesthetic quality is formulated in this work. A new data-set of image pairs with relative labels is constructed by carefully selecting images from the popular AVA data-set. Unlike in aesthetics classification, there is no single threshold which would determine the ranking order of the images across the entire data-set.
This problem is attempted using a deep neural network based approach that is trained on image pairs by incorporating principles from relative learning. Results show that such relative training procedure allows the network to rank the images with a higher accuracy than a state-of-art network trained on the same set of images using binary labels. Further analyzing the results show that training a model using the image pairs learnt better aesthetic features than training on same number of individual binary labelled images.
Additionally, an attempt is made at enhancing the performance of the system by incorporating saliency related information. Given an image, humans might fixate their vision on particular parts of the image, which they might be subconsciously intrigued to. I therefore tried to utilize the saliency information both stand-alone as well as in combination with the global and local aesthetic features by performing two separate sets of experiments. In both the cases, a standard saliency model is chosen and the generated saliency maps are convoluted with the images prior to passing them to the network, thus giving higher importance to the salient regions as compared to the remaining. Thus generated saliency-images are either used independently or along with the global and the local features to train the network. Empirical results show that the saliency related aesthetic features might already be learnt by the network as a sub-set of the global features from automatic feature extraction, thus proving the redundancy of the additional saliency module.
This problem is attempted using a deep neural network based approach that is trained on image pairs by incorporating principles from relative learning. Results show that such relative training procedure allows the network to rank the images with a higher accuracy than a state-of-art network trained on the same set of images using binary labels. Further analyzing the results show that training a model using the image pairs learnt better aesthetic features than training on same number of individual binary labelled images.
Additionally, an attempt is made at enhancing the performance of the system by incorporating saliency related information. Given an image, humans might fixate their vision on particular parts of the image, which they might be subconsciously intrigued to. I therefore tried to utilize the saliency information both stand-alone as well as in combination with the global and local aesthetic features by performing two separate sets of experiments. In both the cases, a standard saliency model is chosen and the generated saliency maps are convoluted with the images prior to passing them to the network, thus giving higher importance to the salient regions as compared to the remaining. Thus generated saliency-images are either used independently or along with the global and the local features to train the network. Empirical results show that the saliency related aesthetic features might already be learnt by the network as a sub-set of the global features from automatic feature extraction, thus proving the redundancy of the additional saliency module.
ContributorsGattupalli, Jaya Vijetha (Author) / Li, Baoxin (Thesis advisor) / Davulcu, Hasan (Committee member) / Liang, Jianming (Committee member) / Arizona State University (Publisher)
Created2016
Description
This research seeks to answer the question if there is a singular relationship between stishovite nucleation and the atomistic structure of the preshocked amorphous SiO$_2$. To do this a stishovite manufacturing method is developed in which 1,152 samples were produced. The majority of these samples did crystallize. The method was produced through two rounds of experiments and fine-tuning with the pressure damp, temperature damp, shock pressure using an NPHug fix, and sample origin. A new random atomic insertion method was used to generate a new and different SiO$_2$ amorphous structure not before seen within the research literature. The optimal values for shock were found to be 60~GPa for randomly atom insertion samples and 55~GPa for quartz origin samples. Temperature damp appeared to have a slight effect optimizing at 0.05~ps and the pressure damp had no visible effect, testing was done with temperature damp from 0.05 to 0.5~ps and pressure damp from 0.1 to 10.0~ps. There appeared to be significant randomness in crystallization behavior. The preshocked and postnucleated samples were transformed into Gaussian fields of crystal, mass, and charge. These fields were divided and classified using a cut-off method taking the number of crystals produced in portions of each simulation and classifying each potion as nucleated or non-nucleated. Data in which some nucleation but not a critical amount was present was removed constituting 2.6\% to 20.3\% of data in all tests. A max method was also used which takes only the maximum portions of each simulation to classify as nucleating. There are three other variables tested within this work, a sample size of 18,000 or 72,728~atoms, Gaussian variance of 1 or 4~\AA, and Convolutional neural network (CNN) architecture of a garden verity or all convolution along with the portioning classification method, sample origination, and Gaussian field type. In total 64 tests were performed to try every combination of variable. No significant classifications were made by the CNNs to nucleation or non-nucleation portions. The results clearly confirmed that the data was not abstracting to atomistic structure and was random by all classifications of the CNNs. The all convolution CNN testing did show smoother outcomes in training with less fluctuations. 59\% of all validation accuracy was held at 0.5 for a random state and 84\% was within $\pm0.02$ of 0.5. It is conclusive that prenucleation structure is not the sole predictor of nucleation behavior. It is not conclusive if prenucleation structure is a partial or non-factor within nucleation of stishovite from amorphous SiO$_2$.
ContributorsChristen, Jonathan Scorr (Author) / Oswald, Jay (Thesis advisor) / Muhich, Christopher (Committee member) / Zhuang, Houlong (Committee member) / Arizona State University (Publisher)
Created2021