Sebastian Cygert - Publications - Bridge of Knowledge

Search

Filters

total: 36

  • Category
  • Year
  • Options

clear Chosen catalog filters disabled

Catalog Publications

Year 2024
  • Adapt Your Teacher: Improving Knowledge Distillation for Exemplar-free Continual Learning
    Publication
    • F. Szatkowski
    • M. Pyła
    • M. Przewięźlikowski
    • S. Cygert
    • B. Twardowski
    • T. Trzciński

    - Year 2024

    In this work, we investigate exemplar-free class incremental learning (CIL) with knowledge distillation (KD) as a regularization strategy, aiming to prevent forgetting. KDbased methods are successfully used in CIL, but they often struggle to regularize the model without access to exemplars of the training data from previous tasks. Our analysis reveals that this issue originates from substantial representation shifts in the teacher...

    Full text to download in external service

  • Category Adaptation Meets Projected Distillation in Generalized Continual Category Discovery
    Publication
    • G. Rypeść
    • D. Marczak
    • S. Cygert
    • T. Trzciński
    • B. Twardowski

    - Year 2024

    "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...

    Full text to download in external service

  • Deep Learning-Based, Multiclass Approach to Cancer Classification on Liquid Biopsy Data
    Publication

    - IEEE Journal of Translational Engineering in Health and Medicine-JTEHM - Year 2024

    Full text to download in external service

  • Deep Learning-Based, Multiclass Approach to Cancer Classification on Liquid Biopsy Data
    Publication

    - IEEE Journal of Translational Engineering in Health and Medicine-JTEHM - Year 2024

    The field of cancer diagnostics has been revolutionized by liquid biopsies, which offer a bridge between laboratory research and clinical settings. These tests are less invasive than traditional biopsies and more convenient than routine imaging methods. Liquid biopsies allow studying of tumor-derived markers in bodily fluids, enabling the development of more precise cancer diagnostic tests for screening, disease monitoring, and...

    Full text available to download

  • Divide and not forget: Ensemble of selectively trained experts in Continual Learning
    Publication
    • G. Rypeść
    • S. Cygert
    • V. Khan
    • T. Trzciński
    • B. Zieliński
    • B. Twardowski

    - Year 2024

    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,...

    Full text available to download

  • Improving platelet‐RNA‐based diagnostics: a comparative analysis of machine learning models for cancer detection and multiclass classification
    Publication

    - Molecular Oncology - Year 2024

    Liquid biopsy demonstrates excellent potential in patient management by providing a minimally invasive and cost-effective approach to detecting and monitoring cancer, even at its early stages. Due to the complexity of liquid biopsy data, machine-learning techniques are increasingly gaining attention in sample analysis, especially for multidimensional data such as RNA expression profiles. Yet, there is no agreement in the community...

    Full text available to download

  • Looking through the past: better knowledge retention for generative replay in continual learning
    Publication
    • V. Khan
    • S. Cygert
    • K. Deja
    • T. Trzciński
    • B. Twardowski

    - IEEE Access - Year 2024

    In this work, we improve the generative replay in a continual learning setting to perform well on challenging scenarios. Because of the growing complexity of continual learning tasks, it is becoming more popular, to apply the generative replay technique in the feature space instead of image space. Nevertheless, such an approach does not come without limitations. In particular, we notice the degradation of the continually trained...

    Full text available to download

  • MagMax: Leveraging Model Merging for Seamless Continual Learning
    Publication
    • D. Marczak
    • B. Twardowski
    • T. Trzciński
    • S. Cygert

    - Year 2024

    This paper introduces a continual learning approach named MagMax, which utilizes model merging to enable large pre-trained models to continuously learn from new data without forgetting previously acquired knowledge. Distinct from traditional continual learning methods that aim to reduce forgetting during task training, MagMax combines sequential fine-tuning with a maximum magnitude weight selection for effective knowledge integration...

    Full text to download in external service

  • Revisiting Supervision for Continual Representation Learning
    Publication
    • D. Marczak
    • S. Cygert
    • T. Trzciński
    • B. Twardowski

    - Year 2024

    "In the field of continual learning, models are designed to learn tasks one after the other. While most research has centered on supervised continual learning, there is a growing interest in unsupervised continual learning, which makes use of the vast amounts of unlabeled data. Recent studies have highlighted the strengths of unsupervised methods, particularly self-supervised learning, in providing robust representations. The improved...

    Full text to download in external service

  • Task-recency bias strikes back: Adapting covariances in Exemplar-Free Class Incremental Learning
    Publication
    • G. Rypeść
    • S. Cygert
    • T. Trzciński
    • B. Twardowski

    - Year 2024

    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...

    Full text available to download

Year 2023
  • Cross-Lingual Knowledge Distillation via Flow-Based Voice Conversion for Robust Polyglot Text-to-Speech
    Publication
    • D. Piotrowski
    • R. Korzeniowski
    • A. Falai
    • S. Cygert
    • K. Pokora
    • G. Tinchev
    • Z. Zhang
    • K. Yanagisawa

    - Year 2023

    In this work, we introduce a framework for cross-lingual speech synthesis, which involves an upstream Voice Conversion (VC) model and a downstream Text-To-Speech (TTS) model. The proposed framework consists of 4 stages. In the first two stages, we use a VC model to convert utterances in the target locale to the voice of the target speaker. In the third stage, the converted data is combined with the linguistic features and durations...

    Full text to download in external service

  • Platelet RNA Sequencing Data Through the Lens of Machine Learning
    Publication

    - Cancers - Year 2023

    Liquid biopsies offer minimally invasive diagnosis and monitoring of cancer disease. This biosource is often analyzed using sequencing, which generates highly complex data that can be used using machine learning tools. Nevertheless, validating the clinical applications of such methods is challenging. It requires: (a) using data from many patients; (b) verifying potential bias concerning sample collection; and (c) adding interpretability...

    Full text available to download

Year 2022
  • Creating new voices using normalizing flows
    Publication
    • P. Biliński
    • T. Merritt
    • A. Ezzerg
    • K. Pokora
    • S. Cygert
    • K. Yanagisawa
    • R. Barra-Chicote
    • D. Korzekwa

    - Year 2022

    Creating realistic and natural-sounding synthetic speech remains a big challenge for voice identities unseen during training. As there is growing interest in synthesizing voices of new speakers, here we investigate the ability of normalizing flows in text-to-speech (TTS) and voice conversion (VC) modes to extrapolate from speakers observed during training to create unseen speaker identities. Firstly, we create an approach for TTS...

    Full text available to download

  • Robust and Efficient Machine Learning Algorithms for Visual Recognition
    Publication

    - Year 2022

    In visual recognition, the task is to identify and localize all objects of interest in the input image. With the ubiquitous presence of visual data in modern days, the role of object recognition algorithms is becoming more significant than ever and ranges from autonomous driving to computer-aided diagnosis in medicine. Current models for visual recognition are dominated by models based on Convolutional Neural Networks (CNNs), which...

    Full text available to download

  • Robust Object Detection with Multi-input Multi-output Faster R-CNN
    Publication

    - Year 2022

    Recent years have seen impressive progress in visual recognition on many benchmarks, however, generalization to the out-of-distribution setting remains a significant challenge. A state-of-the-art method for robust visual recognition is model ensembling. However, recently it was shown that similarly competitive results could be achieved with a much smaller cost, by using multi-input multi-output architecture (MIMO). In this work,...

    Full text to download in external service

  • Robust Object Detection with Multi-input Multi-output Faster R-CNN
    Publication

    - Year 2022

    Recent years have seen impressive progress in visual recognition on many benchmarks, however, generalization to the out-of-distribution setting remains a significant challenge. A state-of-the-art method for robust visual recognition is model ensembling. However, recently it was shown that similarly competitive results could be achieved with a much smaller cost, by using multi-input multi-output architecture (MIMO). In this work,...

    Full text available to download

Year 2021
  • Closer Look at the Uncertainty Estimation in Semantic Segmentation under Distributional Shift

    While recent computer vision algorithms achieve impressive performance on many benchmarks, they lack robustness - presented with an image from a different distribution, (e.g. weather or lighting conditions not considered during training), they may produce an erroneous prediction. Therefore, it is desired that such a model will be able to reliably predict its confidence measure. In this work, uncertainty estimation for the task...

    Full text available to download

  • Robustness in Compressed Neural Networks for Object Detection
    Publication

    - Year 2021

    Model compression techniques allow to significantly reduce the computational cost associated with data processing by deep neural networks with only a minor decrease in average accuracy. Simultaneously, reducing the model size may have a large effect on noisy cases or objects belonging to less frequent classes. It is a crucial problem from the perspective of the models' safety, especially for object detection in the autonomous driving...

    Full text available to download

  • Towards Cancer Patients Classification Using Liquid Biopsy

    Liquid biopsy is a useful, minimally invasive diagnostic and monitoring tool for cancer disease. Yet, developing accurate methods, given the potentially large number of input features, and usually small datasets size remains very challenging. Recently, a novel feature parameterization based on the RNA-sequenced platelet data which uses the biological knowledge from the Kyoto Encyclopedia of Genes and Genomes, combined with a classifier...

    Full text to download in external service

Year 2020
  • Evaluating calibration and robustness of pedestrian detectors
    Publication

    - Year 2020

    In this work robustness and calibration of modern pedestrian detectors are evaluated. Pedestrian detection is a crucial perception com- ponent in autonomous driving and here we study its performance under different image corruptions. Furthermore, we provide analysis of classifi- cation calibration of pedestrian detectors and we show a positive effect of using style-transfer augmentation technique. Our analysis is aimed as a step...

    Full text to download in external service

  • Evaluating Calibration and Robustness of Pedestrian Detectors
    Publication

    - Year 2020

    Full text to download in external service

  • Toward Robust Pedestrian Detection With Data Augmentation
    Publication

    In this article, the problem of creating a safe pedestrian detection model that can operate in the real world is tackled. While recent advances have led to significantly improved detection accuracy on various benchmarks, existing deep learning models are vulnerable to invisible to the human eye changes in the input image which raises concerns about its safety. A popular and simple technique for improving robustness is using data...

    Full text available to download

  • Vehicle Detection with Self-Training for Adaptative Video Processing Embedded Platform
    Publication

    Traffic monitoring from closed-circuit television (CCTV) cameras on embedded systems is the subject of the performed experiments. Solving this problem encounters difficulties related to the hardware limitations, and possible camera placement in various positions which affects the system performance. To satisfy the hardware requirements, vehicle detection is performed using a lightweight Convolutional Neural Network (CNN), named...

    Full text available to download

Year 2019
Year 2018
Year 2017
  • Shape-Based Pose Estimation of Robotic Surgical Instruments
    Publication

    - Year 2017

    We describe a detector of robotic instrument parts in image-guided surgery. The detector consists of a huge ensemble of scale-variant and pose-dedicated, rigid appearance templates. The templates, which are equipped with pose-related keypoints and segmentation masks, allow for explicit pose estimation and segmentation of multiple end-effectors as well as fine-grained non-maximum suppression. We train the templates by grouping examples...

    Full text to download in external service

Year 2014

seen 1858 times