Large multimodal models (LMMs) combine unimodal encoders and large language models (LLMs) to perform multimodal tasks.

Large multimodal models (LMMs) combine unimodal encoders and large language models (LLMs) to perform multimodal tasks. Despite recent advancements towards the interpretability of these models, understanding internal representations of LMMs remains largely a mystery. In this paper, we present a novel framework for the interpretation of LMMs. We propose a dictionary learning based approach, applied to the representation of tokens. The elements of the learned dictionary correspond to our proposed concepts. We show that these concepts are well semantically grounded in both vision and text. Thus we refer to these as "multi-modal concepts". We qualitatively and quantitatively evaluate the results of the learnt concepts. We show that the extracted multimodal concepts are useful to interpret representations of test samples. Finally, we evaluate the disentanglement between different concepts and the quality of grounding concepts visually and textually. We will publicly release our code.

Low-rank approximations of data matrices have become an important tool in Machine Learning in the field of bio-informatics, computer vision, text processing, recommender systems, and others. They allow for embedding high dimensional data in lower dimensional spaces which mitigate effects due to noise, uncover latent relations, or facilitate further processing. In general, MF is a process to find two factor matrices, P R, k m and Q R, k n, to describe a given m-by-n training matrix R in which some entries may be missing. MF can be found in many applications, but we only use collaborative filtering in recommender systems as examples. It is based on the assumption that the entries of R are the historical users' preferences for merchandises, and the task on hand is to predict unobserved user behavior (i.e., missing entries in R) to make a suitable recommendation. In this blog, I discuss about different types of matrix factorization techniques for real-time recommendation engines and their corresponding Python libraries.

Non-negative Matrix Factorization (NMF) is a popular clustering and dimension reduction method by decomposing a non-negative matrix into the product of two lower dimension matrices composed of basis vectors. In this paper, we propose a semi-orthogonal NMF method that enforces one of the matrices to be orthogonal with mixed signs, thereby guarantees the rank of the factorization. Our method preserves strict orthogonality by implementing the Cayley transformation to force the solution path to be exactly on the Stiefel manifold, as opposed to the approximated orthogonality solutions in existing literature. We apply a line search update scheme along with an SVD-based initialization which produces a rapid convergence of the algorithm compared to other existing approaches. In addition, we present formulations of our method to incorporate both continuous and binary design matrices. Through various simulation studies, we show that our model has an advantage over other NMF variations regarding the accuracy of the factorization, rate of convergence, and the degree of orthogonality while being computationally competitive. We also apply our method to a text-mining data on classifying triage notes, and show the effectiveness of our model in reducing classification error compared to the conventional bag-of-words model and other alternative matrix factorization approaches.

Semi-Non-negative Matrix Factorization is a technique that learns a low-dimensional representation of a dataset that lends itself to a clustering interpretation. It is possible that the mapping between this new representation and our original data matrix contains rather complex hierarchical information with implicit lower-level hidden attributes, that classical one level clustering methodologies can not interpret. In this work we propose a novel model, Deep Semi-NMF, that is able to learn such hidden representations that allow themselves to an interpretation of clustering according to different, unknown attributes of a given dataset. We also present a semi-supervised version of the algorithm, named Deep WSF, that allows the use of (partial) prior information for each of the known attributes of a dataset, that allows the model to be used on datasets with mixed attribute knowledge. Finally, we show that our models are able to learn low-dimensional representations that are better suited for clustering, but also classification, outperforming Semi-Non-negative Matrix Factorization, but also other state-of-the-art methodologies variants.