CLIP embeddings have demonstrated remarkable performance across a wide range of multimodal applications. However, these high-dimensional, dense vector representations are not easily interpretable, limiting our understanding of the rich structure of CLIP and its use in downstream applications that require transparency. In this work, we show that the semantic structure of CLIP's latent space can be leveraged to provide interpretability, allowing for the decomposition of representations into semantic concepts. We formulate this problem as one of sparse recovery and propose a novel method, Sparse Linear Concept Embeddings (SpLiCE), for transforming CLIP representations into sparse linear combinations of human-interpretable concepts. Distinct from previous work, \method is task-agnostic and can be used, without training, to explain and even replace traditional dense CLIP representations, maintaining high downstream performance while significantly improving their interpretability. We also demonstrate significant use cases of \method representations including detecting spurious correlations and model editing.

In this work, we show that the semantic structure of CLIP's latent space can be

In this work, we show that the semantic structure of CLIP's latent space can be

CLIP embeddings have demonstrated remarkable performance across a wide range of multimodal applications. However, these high-dimensional, dense vector representations are not easily interpretable, limiting our understanding of the rich structure of CLIP and its use in downstream applications that require transparency. In this work, we show that the semantic structure of CLIP's latent space can be leveraged to provide interpretability, allowing for the decomposition of representations into semantic concepts. We formulate this problem as one of sparse recovery and propose a novel method, Sparse Linear Concept Embeddings (SpLiCE), for transforming CLIP representations into sparse linear combinations of human-interpretable concepts. Distinct from previous work, \method is task-agnostic and can be used, without training, to explain and even replace traditional dense CLIP representations, maintaining high downstream performance while significantly improving their interpretability. We also demonstrate significant use cases of \method representations including detecting spurious correlations and model editing.

Coreset selection, which involves selecting a small subset from an existing training dataset, is an approach to reducing training data, and various approaches have been proposed for this method. In practical situations where these methods are employed, it is often the case that the data distributions differ between the development phase and the deployment phase, with the latter being unknown. Thus, it is challenging to select an effective subset of training data that performs well across all deployment scenarios. We therefore propose Distributionally Robust Coreset Selection (DRCS). DRCS theoretically derives an estimate of the upper bound for the worst-case test error, assuming that the future covariate distribution may deviate within a defined range from the training distribution. Furthermore, by selecting instances in a way that suppresses the estimate of the upper bound for the worst-case test error, DRCS achieves distributionally robust training instance selection. This study is primarily applicable to convex training computation, but we demonstrate that it can also be applied to deep learning under appropriate approximations. In this paper, we focus on covariate shift, a type of data distribution shift, and demonstrate the effectiveness of DRCS through experiments.

Modern applications often leverage multiple views of a subject of study. Within neuroscience, there is growing interest in large-scale simultaneous recordings across multiple brain regions. Understanding the relationship between views (e.g., the neural activity in each region recorded) can reveal fundamental principles about the characteristics of each representation and about the system. However, existing methods to characterize such relationships either lack the expressivity required to capture complex nonlinearities, describe only sources of variance that are shared between views, or discard geometric information that is crucial to interpreting the data. Here, we develop a nonlinear neural network-based method that, given paired samples of high-dimensional views, disentangles low-dimensional shared and private latent variables underlying these views while preserving intrinsic data geometry. Across multiple simulated and real datasets, we demonstrate that our method outperforms competing methods. Using simulated populations of lateral geniculate nucleus (LGN) and V1 neurons we demonstrate our model's ability to discover interpretable shared and private structure across different noise conditions. On a dataset of unrotated and corresponding but randomly rotated MNIST digits, we recover private latents for the rotated view that encode rotation angle regardless of digit class, and places the angle representation on a 1-d manifold, while shared latents encode digit class but not rotation angle. Applying our method to simultaneous Neuropixels recordings of hippocampus and prefrontal cortex while mice run on a linear track, we discover a low-dimensional shared latent space that encodes the animal's position. We propose our approach as a general-purpose method for finding succinct and interpretable descriptions of paired data sets in terms of disentangled shared and private latent variables.

Digital pathology and the integration of artificial intelligence (AI) models have revolutionized histopathology, opening new opportunities. With the increasing availability of Whole Slide Images (WSIs), there's a growing demand for efficient retrieval, processing, and analysis of relevant images from vast biomedical archives. However, processing WSIs presents challenges due to their large size and content complexity. Full computer digestion of WSIs is impractical, and processing all patches individually is prohibitively expensive. In this paper, we propose an unsupervised patching algorithm, Sequential Patching Lattice for Image Classification and Enquiry (SPLICE). This novel approach condenses a histopathology WSI into a compact set of representative patches, forming a "collage" of WSI while minimizing redundancy. SPLICE prioritizes patch quality and uniqueness by sequentially analyzing a WSI and selecting non-redundant representative features. We evaluated SPLICE for search and match applications, demonstrating improved accuracy, reduced computation time, and storage requirements compared to existing state-of-the-art methods. As an unsupervised method, SPLICE effectively reduces storage requirements for representing tissue images by 50%. This reduction enables numerous algorithms in computational pathology to operate much more efficiently, paving the way for accelerated adoption of digital pathology.