Nikolaos Giakoumoglou

News

Research Interests

• Computer Vision
• Deep Learning
• Self-supervised Learning

Education

Teaching Experience

Featured Papers

1. Under Review
   SynCo: Synthetic Hard Negatives for Contrastive Visual Representation Learning

   N. Giakoumoglou and T. Stathaki

   Under review

   Abstract Bib arXiv PDF Code

   Contrastive learning has become a dominant approach in self-supervised visual representation learning, but efficiently leveraging hard negatives, which are samples closely resembling the anchor, remains challenging. We introduce SynCo (Synthetic negatives in Contrastive learning), a novel approach that improves model performance by generating synthetic hard negatives in the representation space. Building on the MoCo framework, SynCo introduces six strategies for creating diverse synthetic hard negatives "on-the-fly" with minimal computational overhead. SynCo achieves faster training and strong representation learning, surpassing MoCo-v2 by +0.4% and MoCHI by +1.0% on ImageNet ILSVRC-2012 linear evaluation. It also transfers more effectively to detection tasks, achieving strong results on PASCAL VOC detection (57.2% AP) and significantly improving over MoCo-v2 on COCO detection (+1.0% AP^bb) and instance segmentation (+0.8% AP^msk).

   @misc{giakoumoglou2024synco, title={{SynCo: Synthetic Hard Negatives for Contrastive Visual Representation Learning}}, author={Nikolaos Giakoumoglou and Tania Stathaki}, year={2024}, eprint={2410.02401}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2410.02401}, }
2. Under Review
   SynCo-v2: An Empirical Study of Training Self-Supervised Vision Transformers with Synthetic Hard Negatives

   N. Giakoumoglou, A. Floros, K. M. Papadopoulos, T. Stathaki

   Under review

   Abstract Bib PDF Code

   We introduce SynCo-v2, a method that integrates synthetic hard negatives into unsupervised vision transformer pretraining to improve representation quality. Our approach is thoroughly benchmarked on ImageNet and transfer learning, image retrieval, copy detection, and image and video segmentation tasks. Notably, our proposed negatives give rise to emergent properties, where learned representations contain explicit information about the semantic content of an image and serve as excellent classifiers (up to +11.3% over baselines). SynCo-v2 achieves these benefits through simple modifications to existing contrastive frameworks and outperforms competing methods while being more resource efficient, e.g., our ViT-B surpasses V-JEPA with ViT-L.

   @article{giakoumoglou2024syncov2, title={SynCo-v2: An Empirical Study of Training Self-Supervised Vision Transformers with Synthetic Hard Negatives}, author={Giakoumoglou, Nikolaos and Floros, Andreas and Papadopoulos, Kleanthis Marios and Stathaki, Tania}, year={2025} }
3. Under Review
   Relational Representation Distillation

   N. Giakoumoglou and T. Stathaki

   Under review

   Abstract Bib arXiv PDF Code

   Knowledge distillation transfers knowledge from large, high-capacity teacher models to more compact student networks. We propose Relational Representation Distillation (RRD) that preserves the relative relationships among instances rather than enforcing absolute separation. Our method introduces separate temperature parameters for teacher and student distributions, creating an implicit information bottleneck that preserves fine-grained relational structure while avoiding the over-separation characteristic of contrastive losses. We establish theoretical connections showing that InfoNCE emerges as a limiting case of our objective when the teacher temperature approaches 0, and empirically demonstrate superior relational alignment and generalization across classification and detection tasks.

   @misc{giakoumoglou2024rrd, title={{Relational Representation Distillation}}, author={Nikolaos Giakoumoglou and Tania Stathaki}, year={2024}, eprint={2407.12073}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2407.12073}, }
4. Under Review
   SNAP: Synthetically Negative Augmented Pretraining for Vision-Language Models

   N. Giakoumoglou, P. Giakoumoglou, K. M. Papadopoulos, A. Floros, T. Stathaki

   Under review

   Abstract Bib PDF

   Vision-language contrastive pretraining relies on large batches of randomly sampled image-text pairs to provide negative examples. We identify two failure modes when generating synthetic hard negatives in the representation space for vision-language models: cross-modal synthetic negatives fall into the modality gap and are trivially rejected, while intra-modal negatives involving the positive pair suffer from positive leakage that sends contradictory gradients. We propose SNAP, which generates intra-modal hard negatives that never involve the positive from either modality, avoiding both failure modes entirely. SNAP is model-agnostic, requires no external generative models, and adds less than 9% training time overhead.

   @article{giakoumoglou2025snap, title={SNAP: Synthetically Negative Augmented Pretraining for Vision-Language Models}, author={Giakoumoglou, Nikolaos and Giakoumoglou, Paschalis and Papadopoulos, Kleanthis Marios and Floros, Andreas and Stathaki, Tania}, year={2026} }

Accepted Papers for Publication

2026

1. ISBI
   Expert Clustering and Knowledge Transfer for Whole Slide Image Classification

   K. M. Papadopoulos, N. Giakoumoglou, A. Floros, P. L. Dragotti, T. Stathaki

   In ISBI (Oral), 2026

   Abstract PDF Code

   Multiple Instance Learning (MIL) is widely adopted for Whole Slide Image (WSI) classification. Existing MIL methods suffer from representation bottlenecks where slide-level aggregation compresses diverse patch information, limiting performance. Our proposed Divide-and-Distill (D&D) framework addresses this by partitioning the feature space into representation-coherent clusters, training specialized expert models on each cluster, and distilling their collective knowledge into a unified model. Experiments across three datasets and six MIL methods demonstrate consistent performance gains without added inference cost.
2. VISAPP
   A Multimodal Approach for Cross-Domain Image Retrieval

   L. Iijima, N. Giakoumoglou and T. Stathaki

   In VISAPP, 2026

   SciTePress Abstract Bib

   Cross-Domain Image Retrieval (CDIR) is a challenging task in computer vision, aiming to match images across different visual domains such as sketches, paintings, and photographs. This paper introduces a novel unsupervised approach to CDIR that incorporates textual context by leveraging pre-trained vision-language models. Our method, dubbed as Caption-Matching (CM), uses generated image captions as a domain-agnostic intermediate representation, enabling effective cross-domain similarity computation without the need for labeled data or fine-tuning. We evaluate our method on standard CDIR benchmark datasets, demonstrating state-of-the-art performance in unsupervised settings with improvements of 24.0% on Office-Home and 132.2% on DomainNet over previous methods.

   @conference{iijima2024caption, author={Lucas Iijima and Nikolaos Giakoumoglou and Tania Stathaki}, title={Caption-Matching: A Multimodal Approach for Cross-Domain Image Retrieval}, booktitle={Proceedings of the 21st International Conference on Computer Vision Theory and Applications - Volume 3: VISAPP}, year={2026}, pages={600-607}, publisher={SciTePress}, organization={INSTICC}, doi={10.5220/0014460000004084}, isbn={978-989-758-804-4}, issn={2184-4321}, }

2025

1. NeurIPS
   Mitigating Representation Bottlenecks in Multiple Instance Learning

   K. M. Papadopoulos, N. Giakoumoglou, A. Floros, T. Stathaki

   In NeurIPS Workshop "MedEurIPS", 2025

   OpenReview Abstract Bib PDF

   Multiple Instance Learning (MIL) is widely used for Whole Slide Image classification in computational pathology, yet existing approaches suffer from a representation bottleneck where diverse patch-level features are compressed into a single slide-level embedding. We propose Divide-and-Distill (D&D), which clusters the feature space into coherent regions, trains expert models on each cluster, and distills their knowledge into a unified model. Experiments demonstrate that D&D consistently improves six state-of-the-art MIL methods in both accuracy and AUC while maintaining single-model inference efficiency.

   @inproceedings{papadopoulos2025mitigating, title={{Mitigating Representation Bottlenecks in Multiple Instance Learning}}, author={Papadopoulos, Kleanthis Marios and Giakoumoglou, Nikolaos and Floros, Andreas and Dragotti, Pier Luigi and Stathaki, Tania}, booktitle={Medical Imaging meets NeurIPS Workshop (MedNeurIPS)}, year={2025}, url={https://openreview.net/forum?id=nywAT7N8Do} }
2. ICIP
   Cluster Contrast for Unsupervised Visual Representation Learning

   N. Giakoumoglou, T. Stathaki

   In ICIP, 2025

   IEEE Abstract Bib arXiv PDF

   We introduce Cluster Contrast (CueCo), a novel approach to unsupervised visual representation learning that effectively combines the strengths of contrastive learning and clustering methods. CueCo is designed to simultaneously scatter and align feature representations within the feature space. Our method achieves 91.40% top-1 classification accuracy on CIFAR-10, 68.56% on CIFAR-100, and 78.65% on ImageNet-100 using linear evaluation with a ResNet-18 backbone.

   @inproceedings{giakoumoglou2025cluster, title={{Cluster Contrast for Unsupervised Visual Representation Learning}}, author={Giakoumoglou, Nikolaos and Stathaki, Tania}, booktitle={2025 IEEE International Conference on Image Processing (ICIP)}, pages={133--138}, year={2025}, organization={IEEE} }
3. ICCV
   Training Self-Supervised Vision Transformers with Synthetic Data and Synthetic Hard Negatives

   N. Giakoumoglou, A. Floros, K. M. Papadopoulos, T. Stathaki

   In ICCV Workshop "LIMIT", 2025

   OpenReview Abstract Bib arXiv PDF

   We build on existing self-supervised learning approaches for vision, drawing inspiration from the adage "fake it till you make it". We investigate two forms of "faking it" in vision transformers: leveraging synthetic data from generative models and generating synthetic hard negatives in the representation space. Our framework, dubbed Syn2Co, combines both approaches and evaluates whether synthetically enhanced training can lead to more robust and transferable visual representations on DeiT-S and Swin-T architectures.

   @inproceedings{giakoumoglou2025fake, title={{Fake \& Square: Training Self-Supervised Vision Transformers with Synthetic Data and Synthetic Hard Negatives}}, author={Nikolaos Giakoumoglou and Andreas Floros and Kleanthis Marios Papadopoulos and Tania Stathaki}, booktitle={Representation Learning with Very Limited Resources: When Data, Modalities, Labels, and Computing Resources are Scarce}, year={2025}, url={https://openreview.net/forum?id=TJUfbYKo2c} }
4. CVPR
   Unsupervised Training of Vision Transformers with Synthetic Negatives

   N. Giakoumoglou, A. Floros, K. M. Papadopoulos, T. Stathaki

   In CVPR Workshop "Visual Concepts", 2025

   OpenReview Abstract Bib arXiv PDF SUPPL Code

   We address the neglected potential of hard negative samples in self-supervised learning. Previous works explored synthetic hard negatives but rarely in the context of vision transformers. We build on this observation and integrate synthetic hard negatives to improve vision transformer representation learning. This simple yet effective technique notably improves the discriminative power of learned representations. Our experiments show performance improvements for both DeiT-S and Swin-T architectures.

   @inproceedings{giakoumoglou2025unsupervised, title={{Unsupervised Training of Vision Transformers with Synthetic Negatives}}, author={Nikolaos Giakoumoglou and Andreas Floros and Kleanthis Marios Papadopoulos and Tania Stathaki}, booktitle={Second Workshop on Visual Concepts}, year={2025}, url={https://openreview.net/forum?id=dg8FuaOKnC}, }

Under Review

1. Under Review
   Discriminative and Consistent Representation Distillation

   N. Giakoumoglou and T. Stathaki

   Under review

   Bib arXiv PDF Code

   @misc{giakoumoglou2024dcd, title={{Discriminative and Consistent Representation Distillation}}, author={Nikolaos Giakoumoglou and Tania Stathaki}, year={2024}, eprint={2407.11802}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2407.11802}, }
2. Under Review
   What Makes Pretraining Data Good for Self-Supervised Learning?

   N. Giakoumoglou, A. Floros, K. M. Papadopoulos, T. Stathaki

   Under review
3. Under Review
   Open-World Semantic Segmentation with Sensitivity Modeling

   A. R. Varvarigos, N. Giakoumoglou, T. Stathaki

   Under review
4. Under Review
   A Review on Discriminative Self-supervised Learning Methods

   N. Giakoumoglou, T. Stathaki, A. Gkelias

   Under review

   Bib arXiv PDF

   @misc{giakoumoglou2024review, title={{A Review on Discriminative Self-supervised Learning Methods in Computer Vision}}, author={Nikolaos Giakoumoglou and Tania Stathaki and Athanasios Gkelias}, year={2025}, eprint={2405.04969}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2405.04969}, }
5. Under Review
   A Review on Artificial Intelligence Methods for Plant Disease and Pest Detection

   N. Giakoumoglou, D. Kapetas, K. M. Papadopoulos, P. Christakakis, T. Stathaki, E. M. Pechlivani

   Under review

Positions of Responsibility