CompInterp

The CompInterp approach to interpretability treats weights and data as a unified modality to provide a compositional perspective on model design, analysis, and manipulation. By combining tensor and neural network paradigms, our $\chi$-nets pave the way for inherently interpretable AI without sacrificing performance.

$\chi$-nets are compositional by design, both in how they are built and in the representations they learn. Their architecture enables mathematical guarantees and weight-based subcircuit analysis, grounding interpretability in formal (de)compositions rather than post-hoc activation-based approximations.

We’re currently scaling CompInterp methods to CNNs and transformers by leveraging their specialised low-rank structure. Learn more about it in our latest talk!

news

Apr 05, 2025	We have a website now!
Mar 04, 2025	We are presenting our poster at CoLoRAI (AAAI 2025)!

selected publications

CoLoRAI
Compositionality Unlocks Deep Interpretable Models

Thomas Dooms, Ward Gauderis, Geraint Wiggins, and Jose Antonio Oramas Mogrovejo

In Connecting Low-Rank Representations in AI: At the 39th Annual AAAI Conference on Artificial Intelligence, Nov 2024

Abs arXiv Bib HTML Video Poster Slides

We propose χ-net, an intrinsically interpretable architecture combining the compositional multilinear structure of tensor networks with the expressivity and efficiency of deep neural networks. χ-nets retain equal accuracy compared to their baseline counterparts. Our novel, efficient diagonalisation algorithm, ODT, reveals linear low-rank structure in a multilayer SVHN model. We leverage this toward formal weightbased interpretability and model compression.
@inproceedings{dooms_compositionality_2024, title = {Compositionality {Unlocks} {Deep} {Interpretable} {Models}}, url = {https://openreview.net/forum?id=bXAt5iZ69l}, urldate = {2025-02-17}, booktitle = {Connecting {Low}-{Rank} {Representations} in {AI}: {At} the 39th {Annual} {AAAI} {Conference} on {Artificial} {Intelligence}}, author = {Dooms, Thomas and Gauderis, Ward and Wiggins, Geraint and Mogrovejo, Jose Antonio Oramas}, month = nov, year = {2024}, }