Adín Ramírez Rivera

Patrik Joslin Kenfack, Kamil Sabbagh, Adín Ramírez Rivera et al.

Fairness has become an essential problem in many domains of Machine Learning (ML), such as classification, natural language processing, and Generative Adversarial Networks (GANs). In this research effort, we study the unfairness of GANs. We formally define a new fairness notion for generative models in terms of the distribution of generated samples sharing the same protected attributes (gender, race, etc.). The defined fairness notion (representational fairness) requires the distribution of the sensitive attributes at the test time to be uniform, and, in particular for GAN model, we show that this fairness notion is violated even when the dataset contains equally represented groups, i.e., the generator favors generating one group of samples over the others at the test time. In this work, we shed light on the source of this representation bias in GANs along with a straightforward method to overcome this problem. We first show on two widely used datasets (MNIST, SVHN) that when the norm of the gradient of one group is more important than the other during the discriminator's training, the generator favours sampling data from one group more than the other at test time. We then show that controlling the groups' gradient norm by performing group-wise gradient norm clipping in the discriminator during the training leads to a more fair data generation in terms of representational fairness compared to existing models while preserving the quality of generated samples.

Sandra Robles, Jonathan S. Gómez, Adín Ramírez Rivera et al.

A key ingredient for semi-analytic models (SAMs) of galaxy formation is the mass assembly history of haloes, encoded in a tree structure. The most commonly used method to construct halo merger histories is based on the outcomes of high-resolution, computationally intensive N-body simulations. We show that machine learning (ML) techniques, in particular Generative Adversarial Networks (GANs), are a promising new tool to tackle this problem with a modest computational cost and retaining the best features of merger trees from simulations. We train our GAN model with a limited sample of merger trees from the Evolution and Assembly of GaLaxies and their Environments (EAGLE) simulation suite, constructed using two halo finders-tree builder algorithms: SUBFIND-D-TREES and ROCKSTAR-ConsistentTrees. Our GAN model successfully learns to generate well-constructed merger tree structures with high temporal resolution, and to reproduce the statistical features of the sample of merger trees used for training, when considering up to three variables in the training process. These inputs, whose representations are also learned by our GAN model, are mass of the halo progenitors and the final descendant, progenitor type (main halo or satellite) and distance of a progenitor to that in the main branch. The inclusion of the latter two inputs greatly improves the final learned representation of the halo mass growth history, especially for SUBFIND-like ML trees. When comparing equally sized samples of ML merger trees with those of the EAGLE simulation, we find better agreement for SUBFIND-like ML trees. Finally, our GAN-based framework can be utilised to construct merger histories of low- and intermediate-mass haloes, the most abundant in cosmological simulations.

Thalles Silva, Adín Ramírez Rivera

We present Consistent Assignment of Views over Random Partitions (CARP), a self-supervised clustering method for representation learning of visual features. CARP learns prototypes in an end-to-end online fashion using gradient descent without additional non-differentiable modules to solve the cluster assignment problem. CARP optimizes a new pretext task based on random partitions of prototypes that regularizes the model and enforces consistency between views' assignments. Additionally, our method improves training stability and prevents collapsed solutions in joint-embedding training. Through an extensive evaluation, we demonstrate that CARP's representations are suitable for learning downstream tasks. We evaluate CARP's representations capabilities in 17 datasets across many standard protocols, including linear evaluation, few-shot classification, k-NN, k-means, image retrieval, and copy detection. We compare CARP performance to 11 existing self-supervised methods. We extensively ablate our method and demonstrate that our proposed random partition pretext task improves the quality of the learned representations by devising multiple random classification tasks. In transfer learning tasks, CARP achieves the best performance on average against many SSL methods trained for a longer time.

Bruno Souza, Marius Aasan, Helio Pedrini et al.

The intersection of vision and language is of major interest due to the increased focus on seamless integration between recognition and reasoning. Scene graphs (SGs) have emerged as a useful tool for multimodal image analysis, showing impressive performance in tasks such as Visual Question Answering (VQA). In this work, we demonstrate that despite the effectiveness of scene graphs in VQA tasks, current methods that utilize idealized annotated scene graphs struggle to generalize when using predicted scene graphs extracted from images. To address this issue, we introduce the SelfGraphVQA framework. Our approach extracts a scene graph from an input image using a pre-trained scene graph generator and employs semantically-preserving augmentation with self-supervised techniques. This method improves the utilization of graph representations in VQA tasks by circumventing the need for costly and potentially biased annotated data. By creating alternative views of the extracted graphs through image augmentations, we can learn joint embeddings by optimizing the informational content in their representations using an un-normalized contrastive approach. As we work with SGs, we experiment with three distinct maximization strategies: node-wise, graph-wise, and permutation-equivariant regularization. We empirically showcase the effectiveness of the extracted scene graph for VQA and demonstrate that these approaches enhance overall performance by highlighting the significance of visual information. This offers a more practical solution for VQA tasks that rely on SGs for complex reasoning questions.

Darwin Saire, Adín Ramírez Rivera

The semantic segmentation task aims at dense classification at the pixel-wise level. Deep models exhibited progress in tackling this task. However, one remaining problem with these approaches is the loss of spatial precision, often produced at the segmented objects' boundaries. Our proposed model addresses this problem by providing an internal structure for the feature representations while extracting a global representation that supports the former. To fit the internal structure, during training, we predict a Gaussian Mixture Model from the data, which, merged with the skip connections and the decoding stage, helps avoid wrong inductive biases. Furthermore, our results show that we can improve semantic segmentation by providing both learning representations (global and local) with a clustering behavior and combining them. Finally, we present results demonstrating our advances in Cityscapes and Synthia datasets.

Thalles Santos Silva, Helio Pedrini, Adín Ramírez Rivera

We present Contextualized Local Visual Embeddings (CLoVE), a self-supervised convolutional-based method that learns representations suited for dense prediction tasks. CLoVE deviates from current methods and optimizes a single loss function that operates at the level of contextualized local embeddings learned from output feature maps of convolution neural network (CNN) encoders. To learn contextualized embeddings, CLoVE proposes a normalized mult-head self-attention layer that combines local features from different parts of an image based on similarity. We extensively benchmark CLoVE's pre-trained representations on multiple datasets. CLoVE reaches state-of-the-art performance for CNN-based architectures in 4 dense prediction downstream tasks, including object detection, instance segmentation, keypoint detection, and dense pose estimation.

Thalles Silva, Helio Pedrini, Adín Ramírez Rivera

This paper introduces a novel approach to improving the training stability of self-supervised learning (SSL) methods by leveraging a non-parametric memory of seen concepts. The proposed method involves augmenting a neural network with a memory component to stochastically compare current image views with previously encountered concepts. Additionally, we introduce stochastic memory blocks to regularize training and enforce consistency between image views. We extensively benchmark our method on many vision tasks, such as linear probing, transfer learning, low-shot classification, and image retrieval on many datasets. The experimental results consolidate the effectiveness of the proposed approach in achieving stable SSL training without additional regularizers while learning highly transferable representations and requiring less computing time and resources.

Karim Galliamov, Syed M Ahsan Kazmi, Adil Khan et al.

Concept Bottleneck Models (CBMs) aim to deliver interpretable predictions by routing decisions through a human-understandable concept layer, yet they often suffer reduced accuracy and concept leakage that undermines faithfulness. We introduce an explicit Information Bottleneck regularizer on the concept layer that penalizes $I(X;C)$ while preserving task-relevant information in $I(C;Y)$, encouraging minimal-sufficient concept representations. We derive two practical variants (a variational objective and an entropy-based surrogate) and integrate them into standard CBM training without architectural changes or additional supervision. Evaluated across six CBM families and three benchmarks, the IB-regularized models consistently outperform their vanilla counterparts. Information-plane analyses further corroborate the intended behavior. These results indicate that enforcing a minimal-sufficient concept bottleneck improves both predictive performance and the reliability of concept-level interventions. The proposed regularizer offers a theoretic-grounded, architecture-agnostic path to more faithful and intervenable CBMs, resolving prior evaluation inconsistencies by aligning training protocols and demonstrating robust gains across model families and datasets.

Marius Aasan, Martine Hjelkrem-Tan, Nico Catalano et al.

Vision Transformers rely on fixed patch tokens that ignore the spatial and semantic structure of images. In this work, we introduce an end-to-end differentiable tokenizer that adapts to image content with pixel-level granularity while remaining backward-compatible with existing architectures for retrofitting pretrained models. Our method uses hierarchical model selection with information criteria to provide competitive performance in both image-level classification and dense-prediction tasks, and even supports out-of-the-box raster-to-vector conversion.

Zeinab Dehghani, Koorosh Aslansefat, Adil Khan et al.

Despite recent advancements in Instruct-based Image Editing models for generating high-quality images, they are known as black boxes and a significant barrier to transparency and user trust. To solve this issue, we introduce SMILE (Statistical Model-agnostic Interpretability with Local Explanations), a novel model-agnostic for localized interpretability that provides a visual heatmap to clarify the textual elements' influence on image-generating models. We applied our method to various Instruction-based Image Editing models like Pix2Pix, Image2Image-turbo and Diffusers-Inpaint and showed how our model can improve interpretability and reliability. Also, we use stability, accuracy, fidelity, and consistency metrics to evaluate our method. These findings indicate the exciting potential of model-agnostic interpretability for reliability and trustworthiness in critical applications such as healthcare and autonomous driving while encouraging additional investigation into the significance of interpretability in enhancing dependable image editing models.

Thalles Silva, Helio Pedrini, Adín Ramírez Rivera

We present Self-Organizing Visual Prototypes (SOP), a new training technique for unsupervised visual feature learning. Unlike existing prototypical self-supervised learning (SSL) methods that rely on a single prototype to encode all relevant features of a hidden cluster in the data, we propose the SOP strategy. In this strategy, a prototype is represented by many semantically similar representations, or support embeddings (SEs), each containing a complementary set of features that together better characterize their region in space and maximize training performance. We reaffirm the feasibility of non-parametric SSL by introducing novel non-parametric adaptations of two loss functions that implement the SOP strategy. Notably, we introduce the SOP Masked Image Modeling (SOP-MIM) task, where masked representations are reconstructed from the perspective of multiple non-parametric local SEs. We comprehensively evaluate the representations learned using the SOP strategy on a range of benchmarks, including retrieval, linear evaluation, fine-tuning, and object detection. Our pre-trained encoders achieve state-of-the-art performance on many retrieval benchmarks and demonstrate increasing performance gains with more complex encoders.

Gabriel Y. Arteaga, Marius Aasan, Rwiddhi Chakraborty et al.

Prototypical self-supervised learning methods consistently suffer from partial prototype collapse, where multiple prototypes converge to nearly identical representations. This undermines their central purpose -- providing diverse and informative targets to guide encoders toward rich representations -- and has led practitioners to over-parameterize prototype sets or add ad-hoc regularizers, which mitigate symptoms rather than address the root cause. We empirically trace the collapse to the joint optimization of encoders and prototypes, which encourages a type of shortcut learning: early in training prototypes drift toward redundant representations that minimize loss without necessarily enhancing representation diversity. To break the joint optimization, we introduce a fully decoupled training strategy that learns prototypes and encoders under separate objectives. Concretely, we model prototypes as a Gaussian mixture updated with an online EM-style procedure, independent of the encoder's loss. This simple yet principled decoupling eliminates prototype collapse without explicit regularization and yields consistently diverse prototypes and stronger downstream performance.

Martine Hjelkrem-Tan, Marius Aasan, Gabriel Y. Arteaga et al.

Vision Transformers naturally accommodate sparsity, yet standard tokenization methods confine features to discrete patch grids. This constraint prevents models from fully exploiting sparse regimes, forcing awkward compromises. We propose Subpixel Placement of Tokens (SPoT), a novel tokenization strategy that positions tokens continuously within images, effectively sidestepping grid-based limitations. With our proposed oracle-guided search, we uncover substantial performance gains achievable with ideal subpixel token positioning, drastically reducing the number of tokens necessary for accurate predictions during inference. SPoT provides a new direction for flexible, efficient, and interpretable ViT architectures, redefining sparsity as a strategic advantage rather than an imposed limitation.

Thalles Silva, Adín Ramírez Rivera

We introduce Consistent Assignment for Representation Learning (CARL), an unsupervised learning method to learn visual representations by combining ideas from self-supervised contrastive learning and deep clustering. By viewing contrastive learning from a clustering perspective, CARL learns unsupervised representations by learning a set of general prototypes that serve as energy anchors to enforce different views of a given image to be assigned to the same prototype. Unlike contemporary work on contrastive learning with deep clustering, CARL proposes to learn the set of general prototypes in an online fashion, using gradient descent without the necessity of using non-differentiable algorithms or K-Means to solve the cluster assignment problem. CARL surpasses its competitors in many representations learning benchmarks, including linear evaluation, semi-supervised learning, and transfer learning.

Darwin Saire, Adín Ramírez Rivera

The semantic segmentation (SS) task aims to create a dense classification by labeling at the pixel level each object present on images. Convolutional neural network (CNN) approaches have been widely used, and exhibited the best results in this task. However, the loss of spatial precision on the results is a main drawback that has not been solved. In this work, we propose to use a multi-task approach by complementing the semantic segmentation task with edge detection, semantic contour, and distance transform tasks. We propose that by sharing a common latent space, the complementary tasks can produce more robust representations that can enhance the semantic labels. We explore the influence of contour-based tasks on latent space, as well as their impact on the final results of SS. We demonstrate the effectiveness of learning in a multi-task setting for hourglass models in the Cityscapes, CamVid, and Freiburg Forest datasets by improving the state-of-the-art without any refinement post-processing.

Miguel Rodríguez Santander, Juan Hernández Albarracín, Adín Ramírez Rivera

Deep learning models need large amounts of data for training. In video recognition and classification, significant advances were achieved with the introduction of new large databases. However, the creation of large-databases for training is infeasible in several scenarios. Thus, existing or small collected databases are typically joined and amplified to train these models. Nevertheless, training neural networks on limited data is not straightforward and comes with a set of problems. In this paper, we explore the effects of stacking databases, model initialization, and data amplification techniques when training with limited data on deep learning models' performance. We focused on the problem of Facial Expression Recognition from videos. We performed an extensive study with four databases at a different complexity and nine deep-learning architectures for video classification. We found that (i) complex training sets translate better to more stable test sets when trained with transfer learning and synthetically generated data, but their performance yields a high variance; (ii) training with more detailed data translates to more stable performance on novel scenarios (albeit with lower performance); (iii) merging heterogeneous data is not a straightforward improvement, as the type of augmentation and initialization is crucial; (iv) classical data augmentation cannot fill the holes created by joining largely separated datasets; and (v) inductive biases help to bridge the gap when paired with synthetic data, but this data is not enough when working with standard initialization techniques.

Albina Khusainova, Adil Khan, Adín Ramírez Rivera et al.

The choice of parameter sharing strategy in multilingual machine translation models determines how optimally parameter space is used and hence, directly influences ultimate translation quality. Inspired by linguistic trees that show the degree of relatedness between different languages, the new general approach to parameter sharing in multilingual machine translation was suggested recently. The main idea is to use these expert language hierarchies as a basis for multilingual architecture: the closer two languages are, the more parameters they share. In this work, we test this idea using the Transformer architecture and show that despite the success in previous work there are problems inherent to training such hierarchical models. We demonstrate that in case of carefully chosen training strategy the hierarchical architecture can outperform bilingual models and multilingual models with full parameter sharing.

Juan F. Hernández Albarracín, Adín Ramírez Rivera

Independent components within low-dimensional representations are essential inputs in several downstream tasks, and provide explanations over the observed data. Video-based disentangled factors of variation provide low-dimensional representations that can be identified and used to feed task-specific models. We introduce MTC-VAE, a self-supervised motion-transfer VAE model to disentangle motion and content from videos. Unlike previous work on video content-motion disentanglement, we adopt a chunk-wise modeling approach and take advantage of the motion information contained in spatiotemporal neighborhoods. Our model yields independent per-chunk representations that preserve temporal consistency. Hence, we reconstruct whole videos in a single forward-pass. We extend the ELBO's log-likelihood term and include a Blind Reenactment Loss as an inductive bias to leverage motion disentanglement, under the assumption that swapping motion features yields reenactment between two videos. We evaluate our model with recently-proposed disentanglement metrics and show that it outperforms a variety of methods for video motion-content disentanglement. Experiments on video reenactment show the effectiveness of our disentanglement in the input space where our model outperforms the baselines in reconstruction quality and motion alignment.

Adín Ramírez Rivera, Adil Khan, Imad E. I. Bekkouch et al.

Anomaly detection suffers from unbalanced data since anomalies are quite rare. Synthetically generated anomalies are a solution to such ill or not fully defined data. However, synthesis requires an expressive representation to guarantee the quality of the generated data. In this paper, we propose a two-level hierarchical latent space representation that distills inliers' feature-descriptors (through autoencoders) into more robust representations based on a variational family of distributions (through a variational autoencoder) for zero-shot anomaly generation. From the learned latent distributions, we select those that lie on the outskirts of the training data as synthetic-outlier generators. And, we synthesize from them, i.e., generate negative samples without seen them before, to train binary classifiers. We found that the use of the proposed hierarchical structure for feature distillation and fusion creates robust and general representations that allow us to synthesize pseudo outlier samples. And in turn, train robust binary classifiers for true outlier detection (without the need for actual outliers during training). We demonstrate the performance of our proposal on several benchmarks for anomaly detection.

Rodolfo Quispe, Darwin Ttito, Adín Ramírez Rivera et al.

Crowd scene analysis has received a lot of attention recently due to the wide variety of applications, for instance, forensic science, urban planning, surveillance and security. In this context, a challenging task is known as crowd counting, whose main purpose is to estimate the number of people present in a single image. A Multi-Stream Convolutional Neural Network is developed and evaluated in this work, which receives an image as input and produces a density map that represents the spatial distribution of people in an end-to-end fashion. In order to address complex crowd counting issues, such as extremely unconstrained scale and perspective changes, the network architecture utilizes receptive fields with different size filters for each stream. In addition, we investigate the influence of the two most common fashions on the generation of ground truths and propose a hybrid method based on tiny face detection and scale interpolation. Experiments conducted on two challenging datasets, UCF-CC-50 and ShanghaiTech, demonstrate that using our ground truth generation methods achieves superior results.

Sandra Robles, Jonathan S. Gómez, Adín Ramírez Rivera et al.

Semi-analytic models are best suited to compare galaxy formation and evolution theories with observations. These models rely heavily on halo merger trees, and their realistic features (i.e., no drastic changes on halo mass or jumps on physical locations). Our aim is to provide a new framework for halo merger tree generation that takes advantage of the results of large volume simulations, with a modest computational cost. We treat halo merger tree construction as a matrix generation problem, and propose a Generative Adversarial Network that learns to generate realistic halo merger trees. We evaluate our proposal on merger trees from the EAGLE simulation suite, and show the quality of the generated trees.

Darwin Saire Pilco, Adín Ramírez Rivera

Recently, graph neural networks (GNNs) have proved to be suitable in tasks on unstructured data. Particularly in tasks as community detection, node classification, and link prediction. However, most GNN models still operate with static relationships. We propose the Graph Learning Network (GLN), a simple yet effective process to learn node embeddings and structure prediction functions. Our model uses graph convolutions to propose expected node features, and predict the best structure based on them. We repeat these steps recursively to enhance the prediction and the embeddings.

Albina Khusainova, Adil Khan, Adín Ramírez Rivera

There is a huge imbalance between languages currently spoken and corresponding resources to study them. Most of the attention naturally goes to the "big" languages: those which have the largest presence in terms of media and number of speakers. Other less represented languages sometimes do not even have a good quality corpus to study them. In this paper, we tackle this imbalance by presenting a new set of evaluation resources for Tatar, a language of the Turkic language family which is mainly spoken in Tatarstan Republic, Russia. We present three datasets: Similarity and Relatedness datasets that consist of human scored word pairs and can be used to evaluate semantic models; and Analogies dataset that comprises analogy questions and allows to explore semantic, syntactic, and morphological aspects of language modeling. All three datasets build upon existing datasets for the English language and follow the same structure. However, they are not mere translations. They take into account specifics of the Tatar language and expand beyond the original datasets. We evaluate state-of-the-art word embedding models for two languages using our proposed datasets for Tatar and the original datasets for English and report our findings on performance comparison.