Browsing by Author "Anderson, Chuck, advisor"

Now showing 1 - 5 of 5

Open Access
Automated deep learning architecture design using differentiable architecture search (DARTS)
(Colorado State University. Libraries, 2019) Sharma, Kartikay, author; Anderson, Chuck, advisor; Beveridge, Ross, committee member; Kirby, Michael, committee member
Creating neural networks by hand is a slow trial-and-error based process. Designing new architectures similar to GoogleNet or FractalNets, which use repeated tree-based structures, is highly likely to be inefficient and sub-optimal because of the large number of possibilities for composing such structures. Recently, neural architecture search algorithms have been able to automate the process of architecture design and have often attained state-of-the-art performances on CIFAR-10, ImageNet and Penn Tree Bank datasets. Even though the search time has been reduced to tens of GPU hours from tens of thousands of GPU hours, most search algorithms rely on additional controllers and hypernetworks to generate architecture encoding or predict weights for sampled architectures. These controllers and hypernetworks might require optimal structure when deployed on a new task on a new dataset. And since this is done by hand, the problem of architecture search is not really solved. Differentiable Architecture Search (DARTS) avoids this problem by using gradient descent methods. In this work, the DARTS algorithm is studied under various conditions and search hyperparameters. DARTS is applied to CIFAR-10 to check reproducibility of the original results. It is also tested in a new setting — on the CheXpert dataset — to discover new architectures and is compared to a baseline DenseNet121 model. The architectures searched using DARTS achieve better performance on the validation set than the baseline model.
Open Access
From neuro-inspired attention methods to generative diffusion: applications to weather and climate
(Colorado State University. Libraries, 2024) Stock, Jason, author; Anderson, Chuck, advisor; Ebert-Uphoff, Imme, committee member; Krishnaswamy, Nikhil, committee member; Sreedharan, Sarath, committee member
Machine learning presents new opportunities for addressing the complexities of atmospheric science, where high-dimensional, sparse, and variable data challenge traditional methods. This dissertation introduces a range of algorithms, motivated specifically by the intricacies of weather and climate applications. These challenges complement those that are fundamental in machine learning, such as extracting relevant features, generating high-quality imagery, and providing interpretable model predictions. To this end, we propose methods to integrate adaptive wavelets and spatial attention into neural networks, showing improvements on tasks with limited data. We design a memory-based model of sequential attention to expressively contextualize a subset of image regions. Additionally, we explore transformer models for image translation, with an emphasis on explainability, that overcome the limitations of convolutional networks. Lastly, we discover meaningful long-range dynamics in oscillatory data from an autoregressive generative diffusion model---a very different approach from the current physics-based models. These methods collectively improve predictive performance and deepen our understanding of both the underlying algorithmic and physical processes. The generality of most of these methods is demonstrated on synthetic data and classical vision tasks, but we place a particular emphasis on their impact in weather and climate modeling. Some notable examples include an application to estimate synthetic radar from satellite imagery, predicting the intensity of tropical cyclones, and modeling global climate variability from observational data for intraseasonal predictability. These approaches, however, are flexible and hold potential for adaptation across various application domains and data modalities.
Open Access
Localized anomaly detection via hierarchical integrated activity discovery
(Colorado State University. Libraries, 2014) Chockalingam, Thiyagarajan, author; Rajopadhye, Sanjay, advisor; Anderson, Chuck, advisor; Pasricha, Sudeep, committee member; Bohm, Wim, committee member
With the increasing number and variety of camera installations, unsupervised methods that learn typical activities have become popular for anomaly detection. In this thesis, we consider recent methods based on temporal probabilistic models and improve them in multiple ways. Our contributions are the following: (i) we integrate the low level processing and the temporal activity modeling, showing how this feedback improves the overall quality of the captured information, (ii) we show how the same approach can be taken to do hierarchical multi-camera processing, (iii) we use spatial analysis of the anomalies both to perform local anomaly detection and to frame automatically the detected anomalies. We illustrate the approach on both traffic data and videos coming from a metro station. We also investigate the application of topic models in Brain Computing Interfaces for Mental Task classification. We observe a classification accuracy of up to 68% for four Mental Tasks on individual subjects.
Open Access
Machine learning models applied to storm nowcasting
(Colorado State University. Libraries, 2020) Cuomo, Joaquin M., author; Anderson, Chuck, advisor; Chandrasekar, V., advisor; Pallickara, Sangmi Lee, committee member; Suryanarayanan, Sid, committee member
Weather nowcasting is heavily dependent on the observation and estimation of radar echoes. There are many different types of deployed nowcasting systems, but none of them based on machine learning, even though it has been an active area of research in the last few years. This work sets the basis for considering machine learning models as real alternatives to current methods by proposing different architectures and comparing them against other nowcasting systems, such as DARTS and STEPS. The methods proposed here are based on residual convolutional encoder-decoder architectures, and they reach the state of the art performance and, in certain scenarios, even outperform them. Different experiments are presented on how the model behaves when using recurrent connections, different loss functions, and different prediction lead times.
Open Access
Phishing detection using machine learning
(Colorado State University. Libraries, 2021) Shirazi, Hossein, author; Ray, Indrakshi, advisor; Anderson, Chuck, advisor; Malaiya, Yashwant K., committee member; Wang, Haonan, committee member
Our society, economy, education, critical infrastructure, and other aspects of our life have become largely dependent on cyber technology. Thus, cyber threats now endanger various aspects of our daily life. Phishing attacks, even with sophisticated detection algorithms, are still the top Internet crime by victim count in 2020. Adversaries learn from their previous attempts to (i) improve attacks and lure more victims and (ii) bypass existing detection algorithms to steal user's identities and sensitive information to increase their financial gain. Machine learning appears to be a promising approach for phishing detection and, classification algorithms distinguish between legitimate and phishing websites. While machine learning algorithms have shown promising results, we observe multiple limitations in existing algorithms. Current algorithms do not preserve the privacy of end-users due to inquiring third-party services. There is a lack of enough phishing samples for training machine learning algorithms and, over-represented targets have a bias in existing datasets. Finally, adversarial sampling attacks degrade the performance of detection models. We propose four sets of solutions to address the aforementioned challenges. We first propose a domain-name-based phishing detection solution that focuses solely on the domain name of websites to distinguish phishing websites from legitimate ones. This approach does not use any third-party services and preserves the privacy of end-users. We then propose a fingerprinting algorithm that consists of finding similarities (using both visual and textual characteristics) between a legitimate targeted website and a given suspicious website. This approach addresses the issue of bias towards over-represented samples in the datasets. Finally, we explore the effect of adversarial sampling attacks on phishing detection algorithms in-depth, starting with feature manipulation strategies. Results degrade the performance of the classification algorithm significantly. In the next step, we focus on two goals of improving the performance of classification algorithms by increasing the size of used datasets and making the detection algorithm robust against adversarial sampling attacks using an adversarial autoencoder.