Timezone: »

Boosting Multi-modal Contrastive Learning with Modern Hopfield Networks and InfoLOOB
Andreas Fürst · Elisabeth Rumetshofer · Johannes Lehner · Viet T. Tran · Fei Tang · Hubert Ramsauer · David Kreil · Michael Kopp · Günter Klambauer · Angela Bitto · Sepp Hochreiter

CLIP yielded impressive results on zero-shot transfer learning tasks and is considered as a foundation model like BERT or GPT3. CLIP vision models that have a rich representation are pre-trained using the InfoNCE objective and natural language supervision before they are fine-tuned on particular tasks. Though CLIP excels at zero-shot transfer learning, it suffers from an explaining away problem, that is, it focuses on one or few features, while neglecting other relevant features. This problem is caused by insufficiently extracting the covariance structure in the original multi-modal data. We suggest to use modern Hopfield networks to tackle the problem of explaining away. Their retrieved embeddings have an enriched covariance structure derived from co-occurrences of features in the stored embeddings. However, modern Hopfield networks increase the saturation effect of the InfoNCE objective which hampers learning. We propose to use the InfoLOOB objective to mitigate this saturation effect. We introduce the novel “Contrastive Leave One Out Boost” (CLOOB), which uses modern Hopfield networks for covariance enrichment together with the InfoLOOB objective. In experiments we compare CLOOB to CLIP after pre-training on the Conceptual Captions and the YFCC dataset with respect to their zero-shot transfer learning performance on other datasets. CLOOB consistently outperforms CLIP at zero-shot transfer learning across all considered architectures and datasets.

Author Information

Andreas Fürst (ELLIS Unit / University Linz)
Elisabeth Rumetshofer (ELLIS Unit / University Linz)
Johannes Lehner (Ellis Unit / University Linz)
Viet T. Tran (ELLIS Unit / University Linz)
Viet T. Tran

PhD student @ JKU Linz | interested in Machine Learning + Physics, DIY stuff and climbing

Fei Tang (Here Technologies)
Hubert Ramsauer (LIT AI Lab, Institute for Machine Learning, Johannes Kepler University Linz, Austria)
David Kreil (Institute of Advanced Research in Artificial Intelligence (IARAI) GmbH)
Michael Kopp (Institute of Advanced Research in Artificial Intelligence (IARAI) GmbH)
Günter Klambauer (Johannes Kepler University Linz)
Angela Bitto (JKU)
Sepp Hochreiter (ELLIS Unit / University Linz)

More from the Same Authors