arXiv subjects

Austin J. Brockmeier

Publications and source records attributed to Austin J. Brockmeier.

16 recordsLinked to original sources

arXiv2026

Disentangling Latent Embeddings with Sparse Linear Concept Cones (SLiCC)

Pretrained vision models encode semantic information in latent embedding spaces that are useful for downstream tasks. We hypothesize that this information can be disentangled to separate the content of complex scenes by decomposing each embedding into multiple concept-specific components. We propose Sparse Linear Concept Cones (SLiCC), a supervised dictionary learning approach that estimates a linear synthesis model consisting of sparse, non-negative combinations of groups of dictionary atoms, whose group-wise activity is guided by multi-label information. Each concept-specific component is a non-negative combination of atoms associated with a label and lies in a corresponding concept cone. We introduce an alternating optimization algorithm and an autoencoder formulation of SLiCC. We apply SLiCC to three types of CLIP and DINOv2 embeddings across five datasets. For CLIP, we additionally exploit image-text alignment to enable a self-supervised SLiCC. We also introduce a fully unsupervised autoencoder that enforces group sparsity based on the norm of neuron activations. Quantitative results show that supervised SLiCC improves concept-filtered image retrieval over unfiltered embeddings, including retrieval of finer-grained information not provided during training. Beyond retrieval of individual concepts, the disentangled components enable concept deletion and recombination across images. We further explore conditional image generation using Image-to-Prompt, with qualitative results illustrating retrieval and generation from manipulated concept representations.

cs.CV↗

arXiv2026

Anomaly Detection via Autoencoder Composite Features and NCE

Unsupervised anomaly detection is a challenging task. Autoencoders (AEs) or generative models are often employed to model the data distribution of normal inputs and subsequently identify anomalous, out-of-distribution inputs by high reconstruction error or low likelihood, respectively. However, AEs may generalize and achieve small reconstruction errors on abnormal inputs. We propose a decoupled training approach for anomaly detection that both an AE and a likelihood model trained with noise contrastive estimation (NCE). After training the AE, NCE estimates a probability density function, to serve as the anomaly score, on the joint space of the AE's latent representation combined with features of the reconstruction quality. To further reduce the false negative rate in NCE we systematically varying the reconstruction features to augment the training and optimize the contrastive Gaussian noise distribution. Experimental assessments on multiple benchmark datasets demonstrate that the proposed approach matches the performance of prevalent state-of-the-art anomaly detection algorithms.

cs.LG↗

arXiv2026

Mitigating Reward Hacking in Realistic Scenarios Through Representation Engineering

Reward hacking is a major problem in current production machine learning systems, resulting in costly and dangerous failures. Here we study how to monitor for and mitigate reward hacking in realistic settings, including tool calls and long contexts. We apply a dynamic intervention to steering directions obtained by three different methods: difference-in-means vectors from labeled rollouts, gradient optimization-based vectors, and vectors extracted when models reference an anti-reward-hacking model spec. This dynamic intervention perturbs model representations less than uniform steering vector addition to the residual stream, allowing for more precise edits. We find that we are able to cheaply and significantly reduce reward hacking without correspondingly significant costs to model coherence or capability, and that steering vectors originally extracted from coding environments also generalize to reducing reward hacking in other domains. Our results pave the way towards latent space monitoring and control over realistic production systems.

cs.LG↗

arXiv2026

Dynamic Neural Optimal Transport for Subset Alignment

We study optimal transport between distributions when only a subset of the target population is relevant to the source. The goal is to map the source distribution to a reweighted target distribution whose density ratio with respect to the full target lies in $[0, c]$. This constraint enables the transported distribution to select an appropriate target subpopulation while limiting how strongly any region of the target can be emphasized. We propose a dynamic neural optimal transport framework for this setting using two neural networks. The first parameterizes a time-dependent potential whose spatial gradient defines the velocity field and whose terminal value enforces the density-ratio constraint. The second parameterizes an interpolation between the source distribution and the learned terminal distribution. Unlike prior neural dynamic optimal transport methods, which typically assume a fixed terminal distribution, our formulation learns the terminal distribution within the support of the target. We demonstrate the method on positive-unlabeled learning, unpaired domain translation, and trajectory inference, where meaningful alignments exist between source and target distributions, but direct distribution matching is inappropriate because the target contains additional or irrelevant components

cs.LG↗

arXiv2026

QSMP: finding representative time series subsequences through Quick Shift+Matrix Profile

Finding representative waveforms in long time series has scientific and practical value in many domains, as it enables summarization and visualization of large time series datasets, and downstream tasks like classification and forecasting. We present here QSMP, a method to find representative waveforms in long time series through a density-guided clustering of time series subsequences. Our method makes a novel connection between Quick Shift, a mode-seeking algorithm, and the Matrix Profile, a time series similarity-search data structure, to adapt Quick Shift to the clustering of subsequences in long time series, with a space complexity that is superior to the state-of-the-art method. Our experiments on synthetic and real datasets show that QSMP can be a valuable tool to summarize and visualize long time series by finding representative waveforms.

cs.LG↗

arXiv2026

A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts

We present a systematic benchmark of out-of-distribution (OOD) detection CSFs through a representation-centric lens. Our study spans CNN and ViT backbones, multiple training paradigms, four image-classification source datasets (CIFAR-10, CIFAR-100, SuperCIFAR-100, and TinyImageNet), and OOD datasets grouped into near, mid, and far regimes using CLIP-derived semantic distances. To compare CSFs across these settings, we employ a multiple-comparison-controlled rank pipeline that identifies top cliques of statistically indistinguishable winners under threshold-free ranking metrics (AURC and AUGRC). The main empirical finding is that the competitive detector family depends more on the learned representation than on score design alone. For both CNNs and ViTs, simple probabilistic scores dominate misclassification detection. On CNNs, margin-based scores are strongest in near-OOD regimes, while geometry-aware scores such as NNGuide, fDBD, and CTM become more competitive as shift severity increases. On fine-tuned ViTs, the top cliques are led mainly by reconstruction- and residual-based scores. To interpret these ranking shifts, we analyze the last-layer representation using Neural Collapse (NC) metrics. The resulting picture is consistent across architectures: prototype- and boundary-aware scores become stronger when the representation is more collapsed and better aligned with classifier weights, whereas weaker-collapse regimes favor gradient- and manifold-based scores. Building on these insights, we propose two contributions: a simple PCA-based projection-filtering procedure that improves detector performance, and an approach that uses NC measurements computed from a trained classifier to predict its competitive out-of-distribution detector shortlist, without requiring any additional OOD data.

cs.LG↗

arXiv2026

Patch2Loc: Learning to Localize Patches for Unsupervised Brain Lesion Detection

Detecting brain lesions as abnormalities observed in magnetic resonance imaging (MRI) is essential for diagnosis and treatment. In the search of abnormalities, such as tumors and malformations, radiologists may benefit from computer-aided diagnostics that use computer vision systems trained with machine learning to segment normal tissue from abnormal brain tissue. While supervised learning methods require annotated lesions, we propose a new unsupervised approach (Patch2Loc) that learns from normal patches taken from structural MRI. We train a neural network model to map a patch back to its spatial location within a slice of the brain volume. During inference, abnormal patches are detected by the relatively higher error and/or variance of the location prediction. This generates a heatmap that can be integrated into pixel-wise methods to achieve finer-grained segmentation. We demonstrate the ability of our model to segment abnormal brain tissues by applying our approach to the detection of tumor tissues in MRI on T2-weighted images from BraTS2021 and MSLUB datasets and T1-weighted images from ATLAS and WMH datasets. We show that it outperforms the state-of-the art in unsupervised segmentation. The implementation for this work can be found on our \href{https://github.com/bakerhassan/Patch2Loc}{GitHub page}. This paper has been accepted at AISTATS 2026.

cs.CV↗

arXiv2025

Convolutional Monge Mapping between EEG Datasets to Support Independent Component Labeling

EEG recordings contain rich information about neural activity but are subject to artifacts, noise, and superficial differences due to sensors, amplifiers, and filtering. Independent component analysis and automatic labeling of independent components (ICs) enable artifact removal in EEG pipelines. Convolutional Monge Mapping Normalization (CMMN) is a recent tool used to achieve spectral conformity of EEG signals, which was shown to improve deep neural network approaches for sleep staging. Here we propose a novel extension of the CMMN method with two alternative approaches to computing the source reference spectrum the target signals are mapped to: (1) channel-averaged and $l_1$-normalized barycenter, and (2) a subject-to-subject mapping that finds the source subject with the closest spectrum to the target subject. Notably, our extension yields space-time separable filters that can be used to map between datasets with different numbers of EEG channels. We apply these filters in an IC classification task, and show significant improvement in recognizing brain versus non-brain ICs. Clinical relevance - EEG recordings are used in the diagnosis and monitoring of multiple neuropathologies, including epilepsy and psychosis. While EEG analysis can benefit from automating artifact removal through independent component analysis and labeling, differences in recording equipment and context (the presence of noise from electrical wiring and other devices) may impact the performance of machine learning models, but these differences can be minimized by appropriate spectral normalization through filtering.

cs.LG↗

arXiv2025

Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity

Chain-of-thought (CoT) outputs let us read a model's step-by-step reasoning. Since any long, serial reasoning process must pass through this textual trace, the quality of the CoT is a direct window into what the model is thinking. This visibility could help us spot unsafe or misaligned behavior (monitorability), but only if the CoT is transparent about its internal reasoning (faithfulness). Fully measuring faithfulness is difficult, so researchers often focus on examining the CoT in cases where the model changes its answer after adding a cue to the input. This proxy finds some instances of unfaithfulness but loses information when the model maintains its answer, and does not investigate aspects of reasoning not tied to the cue. We extend these results to a more holistic sense of monitorability by introducing verbosity: whether the CoT lists every factor needed to solve the task. We combine faithfulness and verbosity into a single monitorability score that shows how well the CoT serves as the model's external `working memory', a property that many safety schemes based on CoT monitoring depend on. We evaluate instruction-tuned and reasoning models on BBH, GPQA, and MMLU. Our results show that models can appear faithful yet remain hard to monitor when they leave out key factors, and that monitorability differs sharply across model families. We release our evaluation code using the Inspect library to support reproducible future work.

cs.LG↗

arXiv2025

Weakly Supervised Object Segmentation by Background Conditional Divergence

As a computer vision task, automatic object segmentation remains challenging in specialized image domains without massive labeled data, such as synthetic aperture sonar images, remote sensing, biomedical imaging, etc. In any domain, obtaining pixel-wise segmentation masks is expensive. In this work, we propose a method for training a masking network to perform binary object segmentation using weak supervision in the form of image-wise presence or absence of an object of interest, which provides less information but may be obtained more quickly from manual or automatic labeling. A key step in our method is that the segmented objects can be placed into background-only images to create realistic images of the objects with counterfactual backgrounds. To create a contrast between the original and counterfactual background images, we propose to first cluster the background-only images and then, during learning, create counterfactual images that blend objects segmented from their original source backgrounds to backgrounds chosen from a targeted cluster. One term in the training loss is the divergence between these counterfactual images and the real object images with backgrounds of the target cluster. The other term is a supervised loss for background-only images. While an adversarial critic could provide the divergence, we use sample-based divergences. We conduct experiments on side-scan and synthetic aperture sonar in which our approach succeeds compared to previous unsupervised segmentation baselines that were only tested on natural images. Furthermore, to show generality we extend our experiments to natural images, obtaining reasonable performance with our method that avoids pretrained networks, generative networks, and adversarial critics. The code for this work can be found at \href{GitHub}{https://github.com/bakerhassan/WSOS}.

cs.CV↗

arXiv2024

Contrastive Learning to Fine-Tune Feature Extraction Models for the Visual Cortex

Predicting the neural response to natural images in the visual cortex requires extracting relevant features from the images and relating those feature to the observed responses. In this work, we optimize the feature extraction in order to maximize the information shared between the image features and the neural response across voxels in a given region of interest (ROI) extracted from the BOLD signal measured by fMRI. We adapt contrastive learning (CL) to fine-tune a convolutional neural network, which was pretrained for image classification, such that a mapping of a given image's features are more similar to the corresponding fMRI response than to the responses to other images. We exploit the recently released Natural Scenes Dataset (Allen et al., 2022) as organized for the Algonauts Project (Gifford et al., 2023), which contains the high-resolution fMRI responses of eight subjects to tens of thousands of naturalistic images. We show that CL fine-tuning creates feature extraction models that enable higher encoding accuracy in early visual ROIs as compared to both the pretrained network and a baseline approach that uses a regression loss at the output of the network to tune it for fMRI response encoding. We investigate inter-subject transfer of the CL fine-tuned models, including subjects from another, lower-resolution dataset (Gong et al., 2023). We also pool subjects for fine-tuning to further improve the encoding performance. Finally, we examine the performance of the fine-tuned models on common image classification tasks, explore the landscape of ROI-specific models by applying dimensionality reduction on the Bhattacharya dissimilarity matrix created using the predictions on those tasks (Mao et al., 2024), and investigate lateralization of the processing for early visual ROIs using salience maps of the classifiers built on the CL-tuned models.

cs.CV↗

arXiv2023

DiME: Maximizing Mutual Information by a Difference of Matrix-Based Entropies

We introduce an information-theoretic quantity with similar properties to mutual information that can be estimated from data without making explicit assumptions on the underlying distribution. This quantity is based on a recently proposed matrix-based entropy that uses the eigenvalues of a normalized Gram matrix to compute an estimate of the eigenvalues of an uncentered covariance operator in a reproducing kernel Hilbert space. We show that a difference of matrix-based entropies (DiME) is well suited for problems involving the maximization of mutual information between random variables. While many methods for such tasks can lead to trivial solutions, DiME naturally penalizes such outcomes. We compare DiME to several baseline estimators of mutual information on a toy Gaussian dataset. We provide examples of use cases for DiME, such as latent factor disentanglement and a multiview representation learning problem where DiME is used to learn a shared representation among views with high mutual information.

cs.LG↗

arXiv2022

The Representation Jensen-Rényi Divergence

We introduce a divergence measure between data distributions based on operators in reproducing kernel Hilbert spaces defined by kernels. The empirical estimator of the divergence is computed using the eigenvalues of positive definite Gram matrices that are obtained by evaluating the kernel over pairs of data points. The new measure shares similar properties to Jensen-Shannon divergence. Convergence of the proposed estimators follows from concentration results based on the difference between the ordered spectrum of the Gram matrices and the integral operators associated with the population quantities. The proposed measure of divergence avoids the estimation of the probability distribution underlying the data. Numerical experiments involving comparing distributions and applications to sampling unbalanced data for classification show that the proposed divergence can achieve state of the art results.

cs.LG↗

arXiv2021

Exploring latent networks in resting-state fMRI using voxel-to-voxel causal modeling feature selection

Functional networks characterize the coordinated neural activity observed by functional neuroimaging. The prevalence of different networks during resting state periods provide useful features for predicting the trajectory of neurodegenerative diseases. Techniques for network estimation rely on statistical correlation or dependence between voxels. Due to the large number of voxels, rather than consider the voxel-to-voxel correlations between all voxels, a small set of seed voxels are chosen. Consequently, the network identification may depend on the selected seeds. As an alternative, we propose to fit first-order linear models with sparse priors on the coefficients to model activity across the entire set of cortical grey matter voxels as a linear combination of a smaller subset of voxels. We propose a two-stage algorithm for voxel subset selection that uses different sparsity-inducing regularization approaches to identify subject-specific causally predictive voxels. To reveal the functional networks among these voxels, we then apply independent component analysis (ICA) to model these voxels' signals as a mixture of latent sources each defining a functional network. Based on the inter-subject similarity of the sources' spatial patterns we identify independent sources that are well-matched across subjects but fail to match the independent sources from a group-based ICA. These are resting state networks, common across subjects that group ICA does not reveal. These complementary networks could help to better identify neurodegeneration, a task left for future work.

eess.SP↗

arXiv2021

Shift-invariant waveform learning on epileptic ECoG

Seizure detection algorithms must discriminate abnormal neuronal activity associated with a seizure from normal neural activity in a variety of conditions. Our approach is to seek spatiotemporal waveforms with distinct morphology in electrocorticographic (ECoG) recordings of epileptic patients that are indicative of a subsequent seizure (preictal) versus non-seizure segments (interictal). To find these waveforms we apply a shift-invariant k-means algorithm to segments of spatially filtered signals to learn codebooks of prototypical waveforms. The frequency of the cluster labels from the codebooks is then used to train a binary classifier that predicts the class (preictal or interictal) of a test ECoG segment. We use the Matthews correlation coefficient to evaluate the performance of the classifier and the quality of the codebooks. We found that our method finds recurrent non-sinusoidal waveforms that could be used to build interpretable features for seizure prediction and that are also physiologically meaningful.

cs.LG↗

arXiv2021

Searching for waveforms on spatially-filtered epileptic ECoG

Seizures are one of the defining symptoms in patients with epilepsy, and due to their unannounced occurrence, they can pose a severe risk for the individual that suffers it. New research efforts are showing a promising future for the prediction and preemption of imminent seizures, and with those efforts, a vast and diverse set of features have been proposed for seizure prediction algorithms. However, the data-driven discovery of nonsinusoidal waveforms for seizure prediction is lacking in the literature, which is in stark contrast with recent works that show the close connection between the waveform morphology of neural oscillations and the physiology and pathophysiology of the brain, and especially its use in effectively discriminating between normal and abnormal oscillations in electrocorticographic (ECoG) recordings of epileptic patients. Here, we explore a scalable, energy-guided waveform search strategy on spatially-projected continuous multi-day ECoG data sets. Our work shows that data-driven waveform learning methods have the potential to not only contribute features with predictive power for seizure prediction, but also to facilitate the discovery of oscillatory patterns that could contribute to our understanding of the pathophysiology and etiology of seizures.

cs.LG↗