Recent advances in machine learning, specifically transformer architecture, have led to significant advancements in commercial domains.
Safe deployment of graph neural networks (GNNs) under distribution shift requires models to provide accurate confidence indicators (CI).
In this paper, we propose PAGER (Principled Analysis of Generalization Errors in Regressors), a framework to systematically detect and characterize failures in deep regression models.
In this paper, we explore the potential of Vision-Language Models (VLMs), specifically CLIP, in predicting visual object relationships, which involves interpreting visual features from images into language-based relations.
We argue that augmentations utilized by existing methods are insufficient to handle large distribution shifts, and hence propose a new approach SiSTA, which first fine-tunes a generative model from the source domain using a single-shot target, and then employs novel sampling strategies for curating synthetic target data.
Advances in the expressivity of pretrained models have increased interest in the design of adaptation protocols which enable safe and effective transfer learning.
Overall, our work carefully studies the effectiveness of popular scoring functions in realistic settings and helps to better understand their limitations.
To this end, we introduce Cross-GAN Auditing (xGA) that, given an established "reference" GAN and a newly proposed "client" GAN, jointly identifies intelligible attributes that are either common across both GANs, novel to the client GAN, or missing from the client GAN.
Limited-Angle Computed Tomography (LACT) is a non-destructive evaluation technique used in a variety of applications ranging from security to medicine.
We present a fully automated framework for building object detectors on satellite imagery without requiring any human annotation or intervention.
The problem of adapting models from a source domain using data from any target domain of interest has gained prominence, thanks to the brittle generalization in deep neural networks.
Overall, our work rigorously contextualizes, both empirically and theoretically, the effects of data-centric properties on augmentation strategies and learning paradigms for graph SSL.
While directly fine-tuning (FT) large-scale, pretrained models on task-specific data is well-known to induce strong in-distribution task performance, recent works have demonstrated that different adaptation protocols, such as linear probing (LP) prior to FT, can improve out-of-distribution generalization.
Model agnostic meta-learning algorithms aim to infer priors from several observed tasks that can then be used to adapt to a new task with few examples.
We are interested in estimating the uncertainties of deep neural networks, which play an important role in many scientific and engineering problems.
We focus on the problem of producing well-calibrated out-of-distribution (OOD) detectors, in order to enable safe deployment of medical image classifiers.
With a minimal overhead of storing the subspace basis set for the source data, CATTAn enables unsupervised alignment between source and target data during adaptation.
Our goal in this paper is to exploit heteroscedastic temperature scaling as a calibration strategy for out of distribution (OOD) detection.
To be successful in single source domain generalization, maximizing diversity of synthesized domains has emerged as one of the most effective strategies.
In this paper, we focus on the challenging problem of multi-source zero shot DG (MDG), where labeled training data from multiple source domains is available but with no access to data from the target domain.
Ranked #18 on Domain Generalization on TerraIncognita
In this paper, we develop a Wasserstein autoencoder (WAE) with a hyperspherical prior for multimodal data in the application of inertial confinement fusion.
Anchoring works by first transforming the input into a tuple consisting of an anchor point drawn from a prior distribution, and a combination of the input sample with the anchor using a pretext encoding scheme.
Using the recent population augmentation graph-based analysis of self-supervised learning, we show theoretically that the success of GCL with popular augmentations is bounded by the graph edit distance between different classes.
1 code implementation • 29 Sep 2021 • Alexandros Karargyris, Renato Umeton, Micah J. Sheller, Alejandro Aristizabal, Johnu George, Srini Bala, Daniel J. Beutel, Victor Bittorf, Akshay Chaudhari, Alexander Chowdhury, Cody Coleman, Bala Desinghu, Gregory Diamos, Debo Dutta, Diane Feddema, Grigori Fursin, Junyi Guo, Xinyuan Huang, David Kanter, Satyananda Kashyap, Nicholas Lane, Indranil Mallick, Pietro Mascagni, Virendra Mehta, Vivek Natarajan, Nikola Nikolov, Nicolas Padoy, Gennady Pekhimenko, Vijay Janapa Reddi, G Anthony Reina, Pablo Ribalta, Jacob Rosenthal, Abhishek Singh, Jayaraman J. Thiagarajan, Anna Wuest, Maria Xenochristou, Daguang Xu, Poonam Yadav, Michael Rosenthal, Massimo Loda, Jason M. Johnson, Peter Mattson
Medical AI has tremendous potential to advance healthcare by supporting the evidence-based practice of medicine, personalizing patient treatment, reducing costs, and improving provider and patient experience.
Explanation techniques that synthesize small, interpretable changes to a given image while producing desired changes in the model prediction have become popular for introspecting black-box models.
Complex scientific inquiries rely increasingly upon large and autonomous multiscale simulation campaigns, which fundamentally require similarity metrics to quantify "sufficient'' changes among data and/or configurations.
no code implementations • 19 Apr 2021 • Bogdan Kustowski, Jim A. Gaffney, Brian K. Spears, Gemma J. Anderson, Rushil Anirudh, Peer-Timo Bremer, Jayaraman J. Thiagarajan, Michael K. G. Kruse, Ryan C. Nora
The method described in this paper can be applied to a wide range of problems that require transferring knowledge from simulations to the domain of experiments.
Unsupervised deep learning methods for solving audio restoration problems extensively rely on carefully tailored neural architectures that carry strong inductive biases for defining priors in the time or spectral domain.
With increased interest in adopting AI methods for clinical diagnosis, a vital step towards safe deployment of such tools is to ensure that the models not only produce accurate predictions but also do not generalize to data regimes where the training data provide no meaningful evidence.
Performance analysis has always been an afterthought during the application development process, focusing on application correctness first.
While this deviation may not be exactly known, its broad characterization is specified a priori, in terms of attributes.
Using this framework, we show that space-filling sample designs, such as blue noise and Poisson disk sampling, which optimize spectral properties, outperform random designs in terms of the generalization gap and characterize this gain in a closed-form.
Large-scale numerical simulations are used across many scientific disciplines to facilitate experimental development and provide insights into underlying physical processes, but they come with a significant computational cost.
Through the use of carefully tailored convolutional neural network architectures, a deep image prior (DIP) can be used to obtain pre-images from latent representation encodings.
A crucial aspect of managing a public health crisis is to effectively balance prevention and mitigation strategies, while taking their socio-economic impact into account.
Here we present a new approach to calibrate an agent-based model -- EpiCast -- using a large set of simulation ensembles for different major metropolitan areas of the United States.
With increasing reliance on the outcomes of black-box models in critical applications, post-hoc explainability tools that do not require access to the model internals are often used to enable humans understand and trust these models.
Deep predictive models rely on human supervision in the form of labeled training data.
In this work, we propose Uncertainty Matching GNN (UM-GNN), that is aimed at improving the robustness of GNN models, particularly against poisoning attacks to the graph structure, by leveraging epistemic uncertainties from the message passing framework.
State-of-the-art under-determined audio source separation systems rely on supervised end-end training of carefully tailored neural network architectures operating either in the time or the spectral domain.
Predictive models that accurately emulate complex scientific processes can achieve exponential speed-ups over numerical simulators or experiments, and at the same time provide surrogates for improving the subsequent analysis.
Automated diagnostic assistants in healthcare necessitate accurate AI models that can be trained with limited labeled data, can cope with severe class imbalances and can support simultaneous prediction of multiple disease conditions.
The wide-spread adoption of representation learning technologies in clinical decision making strongly emphasizes the need for characterizing model reliability and enabling rigorous introspection of model behavior.
The hypothesis that sub-network initializations (lottery) exist within the initializations of over-parameterized networks, which when trained in isolation produce highly generalizable models, has led to crucial insights into network initialization and has enabled efficient inferencing.
Neural networks have become very popular in surrogate modeling because of their ability to characterize arbitrary, high dimensional functions in a data driven fashion.
However, PGD is a brittle optimization technique that fails to identify the right projection (or latent vector) when the observation is corrupted, or perturbed even by a small amount.
Exploiting known semantic relationships between fine-grained tasks is critical to the success of recent model agnostic approaches.
The role of uncertainty quantification (UQ) in deep learning has become crucial with growing use of predictive models in high-risk applications.
Calibration error is commonly adopted for evaluating the quality of uncertainty estimators in deep neural networks.
Limited angle CT reconstruction is an under-determined linear inverse problem that requires appropriate regularization techniques to be solved.
There is significant interest in using modern neural networks for scientific applications due to their effectiveness in modeling highly complex, non-linear problems in a data-driven fashion.
We present function preserving projections (FPP), a scalable linear projection technique for discovering interpretable relationships in high-dimensional data.
With rapid adoption of deep learning in critical applications, the question of when and how much to trust these models often arises, which drives the need to quantify the inherent uncertainties.
Weakly supervised instance labeling using only image-level labels, in lieu of expensive fine-grained pixel annotations, is crucial in several applications including medical image analysis.
2 code implementations • 19 Jul 2019 • Shusen Liu, Di Wang, Dan Maljovec, Rushil Anirudh, Jayaraman J. Thiagarajan, Sam Ade Jacobs, Brian C. Van Essen, David Hysom, Jae-Seung Yeom, Jim Gaffney, Luc Peterson, Peter B. Robinson, Harsh Bhatia, Valerio Pascucci, Brian K. Spears, Peer-Timo Bremer
With the rapid adoption of machine learning techniques for large-scale applications in science and engineering comes the convergence of two grand challenges in visualization.
This paper represents a hybrid approach, where we assume simplified data geometry in the form of subspaces, and consider alignment as an auxiliary task to the primary task of maximizing performance on the source.
This paper provides a general framework to study the effect of sampling properties of training data on the generalization error of the learned machine learning (ML) models.
We also investigate the impact of dense connections on the extraction process that encourage feature reuse and better gradient flow.
Solving inverse problems continues to be a central challenge in computer vision.
We present a novel unsupervised domain adaptation (DA) method for cross-domain visual recognition.
State-of-the-art speaker diarization systems utilize knowledge from external data, in the form of a pre-trained distance metric, to effectively determine relative speaker identities to unseen data.
Unsupervised dimension selection is an important problem that seeks to reduce dimensionality of data, while preserving the most useful characteristics.
Techniques for understanding the functioning of complex machine learning models are becoming increasingly popular, not only to improve the validation process, but also to extract new insights about the data via exploratory analysis.
Though deep network embeddings, e. g. DeepWalk, are widely adopted for community discovery, we argue that feature learning with random node attributes, using graph neural networks, can be more effective.
The hypothesis that computational models can be reliable enough to be adopted in prognosis and patient care is revolutionizing healthcare.
Inferencing with network data necessitates the mapping of its nodes into a vector space, where the relationships are preserved.
Sampling one or more effective solutions from large search spaces is a recurring idea in machine learning, and sequential optimization has become a popular solution.
In automatic speech processing systems, speaker diarization is a crucial front-end component to separate segments from different speakers.
We solve this by making successive estimates on the model and the solution in an iterative fashion.
Processing temporal sequences is central to a variety of applications in health care, and in particular multi-channel Electrocardiogram (ECG) is a highly prevalent diagnostic modality that relies on robust sequence modeling.
Furthermore, we introduce a new approach to discover a diverse set of high quality linear projections and show that in practice the information of $k$ linear projections is often jointly encoded in $\sim k$ axis aligned plots.
Third, we propose an efficient estimator to evaluate the space-filling properties of sample designs in arbitrary dimensions and use it to develop an optimization framework to generate high quality space-filling designs.
The classical techniques require measuring projections, called sinograms, from a full 180$^\circ$ view of the object.
Interpretability has emerged as a crucial aspect of building trust in machine learning systems, aimed at providing insights into the working of complex neural networks that are otherwise opaque to a user.
To this end, we develop the DKMO (Deep Kernel Machine Optimization) framework, that creates an ensemble of dense embeddings using Nystrom kernel approximations and utilizes deep learning to generate task-specific representations through the fusion of the embeddings.
With widespread adoption of electronic health records, there is an increased emphasis for predictive models that can effectively deal with clinical time-series data.
To address this, we propose a bootstrapped version of graph convolutional neural networks (G-CNNs) that utilizes an ensemble of weakly trained G-CNNs, and reduce the sensitivity of models on the choice of graph construction.
Kernel fusion is a popular and effective approach for combining multiple features that characterize different aspects of data.
In this paper, we propose the use of quantile analysis to obtain local scale estimates for neighborhood graph construction.
Influential node detection is a central research topic in social network analysis.
We present an approach to model time series data from resting state fMRI for autism spectrum disorder (ASD) severity classification.
In this paper, we take a step in the direction of tackling the problem of interpretability without compromising the model accuracy.
This paper considers the problem of high dimensional signal detection in a large distributed network whose nodes can collaborate with their one-hop neighboring nodes (spatial collaboration).
However, loss functions such as quantile and quantile Huber generalize the symmetric $\ell_1$ and Huber losses to the asymmetric setting, for a fixed quantile parameter.
We propose an algorithm to learn dictionaries and obtain sparse codes when the data reconstruction fidelity is measured using any smooth PLQ cost function.
For case (c), we propose the combined orthogonal matching pursuit algorithm for coefficient recovery and derive the deterministic sparsity threshold under which recovery of the unique, sparsest coefficient vector is possible.
Algorithmic stability and generalization are desirable characteristics for dictionary learning algorithms that aim to build global dictionaries which can efficiently model any test data similar to the training samples.
Descriptors that have diverse forms can be fused into a unified feature space in a principled manner using kernel methods.