Jai Sharma

Christopher Sun, Jai Sharma, Milind Maiti

Computational methods to accelerate natural disaster response include change detection, map alignment, and vision-aided navigation. Current software functions optimally only on near-nadir images, though off-nadir images are often the first sources of information following a natural disaster. The use of off-nadir images for the aforementioned tasks requires the computation of geocentric pose, which is an aerial vehicle's spatial orientation with respect to gravity. This study proposes a deep learning ensemble framework to predict geocentric pose using 5,923 near-nadir and off-nadir RGB satellite images of cities worldwide. First, a U-Net Fully Convolutional Neural Network predicts the pixel-wise above-ground elevation mask of the RGB images. Then, the elevation masks are concatenated with the RGB images to form four-channel inputs fed into a second convolutional model, which predicts orientation angle and magnification scale. A performance accuracy of R2=0.917 significantly outperforms previous methodologies. In addition, outlier removal is performed through supervised interpolation, and a sensitivity analysis of elevation masks is conducted to gauge the usefulness of data features, motivating future avenues of feature engineering. The high-accuracy software built in this study contributes to mapping and navigation procedures for effective disaster response to save lives.

Jai Sharma, Yifan Wang, Bryan Li

Understanding dependencies between variables is critical for interpretability and efficient generation in masked diffusion models (MDMs), yet these models primarily expose marginal conditional distributions and do not explicitly represent inter-variable dependence. We propose a neural framework for estimating pairwise conditional mutual information (MI) directly from the hidden states of a pretrained MDM, using ground-truth MI computed from the model's own conditional distributions for supervision. The resulting estimator captures the model's internal belief about dependency structure and predicts the full MI matrix in a single forward pass, enabling MI-guided parallel decoding by identifying conditionally independent subsets of variables. We evaluate our approach on Sudoku and protein sequence generation with ESM-C, where the MI maps recover known structural constraints and enable a 3-5x magnitude reduction in inference-time forward passes compared to sequential decoding, while preserving generative quality and outperforming entropy-based parallelization methods.

Christopher Sun, Jai Sharma, Milind Maiti

Dropout Regularization, serving to reduce variance, is nearly ubiquitous in Deep Learning models. We explore the relationship between the dropout rate and model complexity by training 2,000 neural networks configured with random combinations of the dropout rate and the number of hidden units in each dense layer, on each of the three data sets we selected. The generated figures, with binary cross entropy loss and binary accuracy on the z-axis, question the common assumption that adding depth to a dense layer while increasing the dropout rate will certainly enhance performance. We also discover a complex correlation between the two hyperparameters that we proceed to quantify by building additional machine learning and Deep Learning models which predict the optimal dropout rate given some hidden units in each dense layer. Linear regression and polynomial logistic regression require the use of arbitrary thresholds to select the cost data points included in the regression and to assign the cost data points a binary classification, respectively. These machine learning models have mediocre performance because their naive nature prevented the modeling of complex decision boundaries. Turning to Deep Learning models, we build neural networks that predict the optimal dropout rate given the number of hidden units in each dense layer, the desired cost, and the desired accuracy of the model. Though, this attempt encounters a mathematical error that can be attributed to the failure of the vertical line test. The ultimate Deep Learning model is a neural network whose decision boundary represents the 2,000 previously generated data points. This final model leads us to devise a promising method for tuning hyperparameters to minimize computational expense yet maximize performance. The strategy can be applied to any model hyperparameters, with the prospect of more efficient tuning in industrial models.