Oriel — AI & technology worth noticing, every day.

Disentangling Latent Embeddings with Sparse Linear Concept Cones (SLiCC)

Paper · https://arxiv.org/abs/2508.20322

arXiv:2508.20322v2 Announce Type: replace Abstract: 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.

First seen by Oriel on .

No saved summary available.

Aliases

SourceExternal IDURL
arXiv AI Papersoai:arXiv.org:2508.20322v2https://arxiv.org/abs/2508.20322 Read discussion