arXiv subjects

Grzegorz Rypeść

Publications and source records attributed to Grzegorz Rypeść.

8 recordsLinked to original sources

arXiv2026

Gradient-Free Continual Learning

Neural networks are notorious for forgetting old skills when taught new ones - a problem known as catastrophic forgetting. Standard continual learning techniques try to fix this by saving old data or relying on complex gradient updates, but these methods fail when past data cannot be stored due to memory or privacy constraints. To solve this, we propose EvoCL, a gradient-free approach that uses evolutionary algorithms to update the network without needing old data or gradients. EvoCL uses a lightweight adapter module to translate saved representations from past tasks into the model's current space, allowing it to learn new tasks while keeping past knowledge intact. Across multiple benchmarks, EvoCL matches or exceeds standard performance under strict memory constraints, offering a simple and flexible new direction for continual learning. The code to reproduce these results is available at https://github.com/grypesc/EvoCL.

cs.LG↗

arXiv2024

Task-recency bias strikes back: Adapting covariances in Exemplar-Free Class Incremental Learning

Exemplar-Free Class Incremental Learning (EFCIL) tackles the problem of training a model on a sequence of tasks without access to past data. Existing state-of-the-art methods represent classes as Gaussian distributions in the feature extractor's latent space, enabling Bayes classification or training the classifier by replaying pseudo features. However, we identify two critical issues that compromise their efficacy when the feature extractor is updated on incremental tasks. First, they do not consider that classes' covariance matrices change and must be adapted after each task. Second, they are susceptible to a task-recency bias caused by dimensionality collapse occurring during training. In this work, we propose AdaGauss -- a novel method that adapts covariance matrices from task to task and mitigates the task-recency bias owing to the additional anti-collapse loss function. AdaGauss yields state-of-the-art results on popular EFCIL benchmarks and datasets when training from scratch or starting from a pre-trained backbone. The code is available at: https://github.com/grypesc/AdaGauss.

cs.LG↗

arXiv2024

Category Adaptation Meets Projected Distillation in Generalized Continual Category Discovery

Generalized Continual Category Discovery (GCCD) tackles learning from sequentially arriving, partially labeled datasets while uncovering new categories. Traditional methods depend on feature distillation to prevent forgetting the old knowledge. However, this strategy restricts the model's ability to adapt and effectively distinguish new categories. To address this, we introduce a novel technique integrating a learnable projector with feature distillation, thus enhancing model adaptability without sacrificing past knowledge. The resulting distribution shift of the previously learned categories is mitigated with the auxiliary category adaptation network. We demonstrate that while each component offers modest benefits individually, their combination - dubbed CAMP (Category Adaptation Meets Projected distillation) - significantly improves the balance between learning new information and retaining old. CAMP exhibits superior performance across several GCCD and Class Incremental Learning scenarios. The code is available at https://github.com/grypesc/CAMP.

cs.LG↗

arXiv2024

ESC: Evolutionary Stitched Camera Calibration in the Wild

This work introduces a novel end-to-end approach for estimating extrinsic parameters of cameras in multi-camera setups on real-life sports fields. We identify the source of significant calibration errors in multi-camera environments and address the limitations of existing calibration methods, particularly the disparity between theoretical models and actual sports field characteristics. We propose the Evolutionary Stitched Camera calibration (ESC) algorithm to bridge this gap. It consists of image segmentation followed by evolutionary optimization of a novel loss function, providing a unified and accurate multi-camera calibration solution with high visual fidelity. The outcome allows the creation of virtual stitched views from multiple video sources, being as important for practical applications as numerical accuracy. We demonstrate the superior performance of our approach compared to state-of-the-art methods across diverse real-life football fields with varying physical characteristics.

cs.CV↗

arXiv2024

Divide and not forget: Ensemble of selectively trained experts in Continual Learning

Class-incremental learning is becoming more popular as it helps models widen their applicability while not forgetting what they already know. A trend in this area is to use a mixture-of-expert technique, where different models work together to solve the task. However, the experts are usually trained all at once using whole task data, which makes them all prone to forgetting and increasing computational burden. To address this limitation, we introduce a novel approach named SEED. SEED selects only one, the most optimal expert for a considered task, and uses data from this task to fine-tune only this expert. For this purpose, each expert represents each class with a Gaussian distribution, and the optimal expert is selected based on the similarity of those distributions. Consequently, SEED increases diversity and heterogeneity within the experts while maintaining the high stability of this ensemble method. The extensive experiments demonstrate that SEED achieves state-of-the-art performance in exemplar-free settings across various scenarios, showing the potential of expert diversification through data in continual learning.

cs.LG↗

arXiv2023

Sports Camera Pose Refinement Using an Evolution Strategy

This paper presents a robust end-to-end method for sports cameras extrinsic parameters optimization using a novel evolution strategy. First, we developed a neural network architecture for an edge or area-based segmentation of a sports field. Secondly, we implemented the evolution strategy, which purpose is to refine extrinsic camera parameters given a single, segmented sports field image. Experimental comparison with state-of-the-art camera pose refinement methods on real-world data demonstrates the superiority of the proposed algorithm. We also perform an ablation study and propose a way to generalize the method to additionally refine the intrinsic camera matrix.

cs.NE↗

arXiv2023

Active Visual Exploration Based on Attention-Map Entropy

Active visual exploration addresses the issue of limited sensor capabilities in real-world scenarios, where successive observations are actively chosen based on the environment. To tackle this problem, we introduce a new technique called Attention-Map Entropy (AME). It leverages the internal uncertainty of the transformer-based model to determine the most informative observations. In contrast to existing solutions, it does not require additional loss components, which simplifies the training. Through experiments, which also mimic retina-like sensors, we show that such simplified training significantly improves the performance of reconstruction, segmentation and classification on publicly available datasets.

cs.CV↗

arXiv2022

Reinforcement Learning for on-line Sequence Transformation

A number of problems in the processing of sound and natural language, as well as in other areas, can be reduced to simultaneously reading an input sequence and writing an output sequence of generally different length. There are well developed methods that produce the output sequence based on the entirely known input. However, efficient methods that enable such transformations on-line do not exist. In this paper we introduce an architecture that learns with reinforcement to make decisions about whether to read a token or write another token. This architecture is able to transform potentially infinite sequences on-line. In an experimental study we compare it with state-of-the-art methods for neural machine translation. While it produces slightly worse translations than Transformer, it outperforms the autoencoder with attention, even though our architecture translates texts on-line thereby solving a more difficult problem than both reference methods.

cs.LG↗