Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 145 for “"distribution shift"”.
-
Tackling Distribution Shift - Detection and Mitigation
… form of an outlier or overall changes in data distribution after model deployment result in a performance drop. Owing to these changes that induce distributional shifts, we propose two methodologies; the first is the detection of these shifts, and the second is adapting the model to overcome …
-
Addressing Distribution Shift in Robotic Imitation Learning
… and actions. Unfortunately, IL is subject to distribution shift: when a robot encounters data not adequately covered by its training dataset, it can fail, sometimes catastrophically. This thesis presents approaches to resolve distribution shift in robotic IL. The first two methods presented …
-
Learning with Distribution Shift: Theory and Practice
… sequentially encounter tasks with distinct distributions. While state-of-the-art machine learning algorithms (e.g., deep neural networks) designed for single-task problems achieve remarkable success in static environments, they suffer from \emph{catastrophic forgetting} in CL scenarios - a …
-
SCALABLE BAYESIAN AND ADAPTIVE LEARNING UNDER DISTRIBUTION SHIFT
… inference to approximate the posterior distribution over model parameters efficiently. This design enables uncertainty-aware inference in high-capacity architectures while significantly reducing the computational and memory overhead associated with full Bayesian methods. In addition, the …
-
Robust and Fair Machine Learning under Distribution Shift
… this assumption is quite restrictive because the distribution shift can exist from the training data to the test data in many scenarios. In addition, the goal of traditional machine learning model is to maximize the prediction performance, e.g., accuracy, based on the historical training data, …
-
Dynamic pricing of airline ancillaries under distribution shift
Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2024-12-01
-
Time series forecasting under distribution shift in environmental applications
Climate change affects all regions of the world. Ice is melting and the sea is rising. In some regions extreme weather events and rainfall are becoming more frequent while others are experiencing higher temperature and droughts. Heavy rain and other extreme weather events are becoming more frequent …
-
Deep Time Series Modeling: From Distribution Regularity to Distribution Shift
… for deep time series modeling from a novel distribution perspective. After in-depth exploration, I categorize and study two essential characteristics of time series, i.e., the distribution regularity and the distribution shift, respectively. I will investigate how can time series data …
-
Mitigating the challenges of distribution shift under strong computational constraints
… methods to address the challenges of distribution shift under strong computational constraints in deep learning vision tasks. Through a series of contributions, it demonstrates the effectiveness of unsupervised learning in enabling efficient and adaptable models suitable for …
-
Addressing Dataset Bias in Android Malware Classifiers under Distribution Shift
… of the threat landscape. In particular, natural distribution shift in malicious and benign app over time leads to significant degradation in detection performance. While prior work has studied mitigation methods, the causes for such drift seen in popular datasets are less explored. This thesis …
-
Investigating Conformal Prediction Under Distribution Shift and Long-tailed Data
… its performance is known to degrade under distribution shift and long-tailed class distributions, which are often present in real world applications. The aim of this thesis is to understand and help mitigate this degradation. We characterize the performance of numerous CP methods under …
-
Using Ensemble Disagreement to Stabilize Conformal Prediction Under Distribution Shift
… with a guaranteed average coverage level for in-distribution data. However, the “vanilla” conformal score can often under-cover in hard or out-of-distribution (OOD) regions under drift. Motivated by difficulty-adaptive conformal approaches, this study investigates normalizing conformal scores by …
-
Distributional robustness: towards challenging and real-world settings of distribution shift
… world we live in causes continuous data distribution shifts for any machine learning application. As machine learning models are notoriously not robust against such distribution shifts, it is of our interest to develop techniques to deal with this problem. <strong>Hence, this thesis …
-
Conformal Inference for Online Anomaly Detection under Distribution Shift in Equity Markets
… tail probabilities, is complicated by extremal distribution shifts that originate from market-wide events or structural changes in a price series. In this thesis, we explore the integration of adaptive conformal inference, as introduced by Gibbs and Candés, into a widely used linear model of …
-
Learning under Dynamics: Distribution Shift Detection and Continual Learning via Modulated Learning
… complementary capabilities: detecting when data shifts beyond known boundaries, and adapting to novel concepts without catastrophically forgetting the past. This dissertation develops four methods that address these capabilities by making architectural choices adaptive where existing approaches …
-
Towards science-guided knowledge distillation for multimodal learning under cross-domain distribution shift
… with extreme label sparsity and persistent distribution shifts arising from evolving sampling devices, diverse data modalities, and dynamic operational characteristics. This dissertation explores these challenges and presents a comprehensive methodology to bridge the adaptability gap by …
-
KLIP: Localized distribution shift detection via KL-divergence with diffusion priors in inverse problems
… as well as some capacity to detect out-of-distribution (OOD) images. However, existing approaches to OOD detection often require some knowledge of the shifted distribution, fail to detect subtle or localized distribution shifts, and operate on full images, rather than the indirect …
-
On machine learning methods for time series with financial applications
… which is entitled “Time Series Prediction under Distribution Shift using Differentiable Forgetting”, proposes a bi-level optimisation framework for updating time series prediction models in response to distribution shift. The third project, “Rethinking Neural Relational Inference for Granger …
-
Bias-variance Trade-off and Uncertainty Quantification: Effects of Data Distribution in Image Classification
… applications. We study the setup where the test distribution may induce a drop in the accuracy of the predictions due to distribution shift. The training of deep neural networks is often regularized either implicitly, for example by early stopping the gradient descent, or explicitly, by adding an …
-
Computational foundations of reliable learning
… assume clean i.i.d. samples from the same distribution on which the classifier is evaluated. Efficient algorithms often assume more: structured marginals, benign noise, no outliers, or stable deployment conditions. These assumptions make learning tractable, but they are also the weak point …
Page 1 of 8