Michał Woźniak

Jan Idziak, Artjoms Šeļa, Michał Woźniak et al.

The paper discusses an approach to decipher large collections of handwritten index cards of historical dictionaries. Our study provides a working solution that reads the cards, and links their lemmas to a searchable list of dictionary entries, for a large historical dictionary entitled the Dictionary of the 17th- and 18th-century Polish, which comprizes 2.8 million index cards. We apply a tailored handwritten text recognition (HTR) solution that involves (1) an optimized detection model; (2) a recognition model to decipher the handwritten content, designed as a spatial transformer network (STN) followed by convolutional neural network (RCNN) with a connectionist temporal classification layer (CTC), trained using a synthetic set of 500,000 generated Polish words of different length; (3) a post-processing step using constrained Word Beam Search (WBC): the predictions were matched against a list of dictionary entries known in advance. Our model achieved the accuracy of 0.881 on the word level, which outperforms the base RCNN model. Within this study we produced a set of 20,000 manually annotated index cards that can be used for future benchmarks and transfer learning HTR applications.

Jędrzej Kozal, Jan Wasilewski, Bartosz Krawczyk et al.

Continual learning poses a fundamental challenge for modern machine learning systems, requiring models to adapt to new tasks while retaining knowledge from previous ones. Addressing this challenge necessitates the development of efficient algorithms capable of learning from data streams and accumulating knowledge over time. This paper proposes a novel approach to continual learning utilizing the weight consolidation method. Our method, a simple yet powerful technique, enhances robustness against catastrophic forgetting by interpolating between old and new model weights after each novel task, effectively merging two models to facilitate exploration of local minima emerging after arrival of new concepts. Moreover, we demonstrate that our approach can complement existing rehearsal-based replay approaches, improving their accuracy and further mitigating the forgetting phenomenon. Additionally, our method provides an intuitive mechanism for controlling the stability-plasticity trade-off. Experimental results showcase the significant performance enhancement to state-of-the-art experience replay algorithms the proposed weight consolidation approach offers. Our algorithm can be downloaded from https://github.com/jedrzejkozal/weight-interpolation-cl.

Jędrzej Kozal, Michał Leś, Paweł Zyblewski et al.

The abundance of information in digital media, which in today's world is the main source of knowledge about current events for the masses, makes it possible to spread disinformation on a larger scale than ever before. Consequently, there is a need to develop novel fake news detection approaches capable of adapting to changing factual contexts and generalizing previously or concurrently acquired knowledge. To deal with this problem, we propose a lifelong learning-inspired approach, which allows for fake news detection in multiple languages and the mutual transfer of knowledge acquired in each of them. Both classical feature extractors, such as Term frequency-inverse document frequency or Latent Dirichlet Allocation, and integrated deep NLP (Natural Language Processing) BERT (Bidirectional Encoder Representations from Transformers) models paired with MLP (Multilayer Perceptron) classifier, were employed. The results of experiments conducted on two datasets dedicated to the fake news classification task (in English and Spanish, respectively), supported by statistical analysis, confirmed that utilization of additional languages could improve performance for traditional methods. Also, in some cases supplementing the deep learning method with classical ones can positively impact obtained results. The ability of models to generalize the knowledge acquired between the analyzed languages was also observed.

Michał Leś, Michał Woźniak

Automatic music transcription (AMT) is one of the most challenging tasks in the music information retrieval domain. It is the process of converting an audio recording of music into a symbolic representation containing information about the notes, chords, and rhythm. Current research in this domain focuses on developing new models based on transformer architecture or using methods to perform semi-supervised training, which gives outstanding results, but the computational cost of training such models is enormous. This work shows how to employ easily generated synthesized audio data produced by software synthesizers to train a universal model. It is a good base for further transfer learning to quickly adapt transcription model for other instruments. Achieved results prove that using synthesized data for training may be a good base for pretraining general-purpose models, where the task of transcription is not focused on one instrument.

Jędrzej Kozal, Michał Woźniak

Since data is the fuel that drives machine learning models, and access to labeled data is generally expensive, semi-supervised methods are constantly popular. They enable the acquisition of large datasets without the need for too many expert labels. This work combines self-labeling techniques with active learning in a selective sampling scenario. We propose a new method that builds an ensemble classifier. Based on an evaluation of the inconsistency of the decisions of the individual base classifiers for a given observation, a decision is made on whether to request a new label or use the self-labeling. In preliminary studies, we show that naive application of self-labeling can harm performance by introducing bias towards selected classes and consequently lead to skewed class distribution. Hence, we also propose mechanisms to reduce this phenomenon. Experimental evaluation shows that the proposed method matches current selective sampling methods or achieves better results.

Szymon Wojciechowski, Michał Woźniak

Many machine learning tasks aim to find models that work well not for a single, but for a group of criteria, often opposing ones. One such example is imbalanced data classification, where, on the one hand, we want to achieve the best possible classification quality for data from the minority class without degrading the classification quality of the majority class. One solution is to propose an aggregate learning criterion and reduce the multi-objective learning task to a single-criteria optimization problem. Unfortunately, such an approach is characterized by ambiguity of interpretation since the value of the aggregated criterion does not indicate the value of the component criteria. Hence, there are more and more proposals for algorithms based on multi-objective optimization (MOO), which can simultaneously optimize multiple criteria. However, such an approach results in a set of multiple non-dominated solutions (Pareto front). The selection of a single solution from the Pareto front is a challenge itself, and much attention is paid to the issue of how to select it considering user preferences, as well as how to compare solutions returned by different MOO algorithms among themselves. Thus, a significant gap has been identified in the classifier evaluation methodology, i.e., how to reliably compare methods returning single solutions with algorithms returning solutions in the form of Pareto fronts. To fill the aforementioned gap, this article proposes a new, reliable way of evaluating algorithms based on multi-objective algorithms with methods that return single solutions while pointing out solutions from a Pareto front tailored to the user's preferences. This work focuses only on algorithm comparison, not their learning. The algorithms selected for this study are illustrative to help understand the proposed approach.

Szymon Wojciechowski, Michał Woźniak

One of the significant problems associated with imbalanced data classification is the lack of reliable metrics. This runs primarily from the fact that for most real-life (as well as commonly used benchmark) problems, we do not have information from the user on the actual form of the loss function that should be minimized. Although it is pretty common to have metrics indicating the classification quality within each class, for the end user, the analysis of several such metrics is then required, which in practice causes difficulty in interpreting the usefulness of a given classifier. Hence, many aggregate metrics have been proposed or adopted for the imbalanced data classification problem, but there is still no consensus on which should be used. An additional disadvantage is their ambiguity and systematic bias toward one class. Moreover, their use in analyzing experimental results in recognition of those classification models that perform well for the chosen aggregated metrics is burdened with the drawbacks mentioned above. Hence, the paper proposes a simple approach to analyzing the popular parametric metric $F_β$. We point out that it is possible to indicate for a given pool of analyzed classifiers when a given model should be preferred depending on user requirements.

Patryk Jasiorski, Marek Klonowski, Michał Woźniak

AI models need to be unlearned to fulfill the requirements of legal acts such as the AI Act or GDPR, and also because of the need to remove toxic content, debiasing, the impact of malicious instances, or changes in the data distribution structure in which a model works. Unfortunately, removing knowledge may cause undesirable side effects, such as a deterioration in model performance. In this paper, we investigate the problem of adversarial unlearning, where a malicious party intentionally sends unlearn requests to deteriorate the model's performance maximally. We show that this phenomenon and the adversary's capabilities depend on many factors, primarily on the backbone model itself and strategy/limitations in selecting data to be unlearned. The main result of this work is a new method of protecting model performance from these side effects, both in the case of unlearned behavior resulting from spontaneous processes and adversary actions.

Jędrzej Kozal, Jan Wasilewski, Alif Ashrafee et al.

Memorization impacts the performance of deep learning algorithms. Prior works have studied memorization primarily in the context of generalization and privacy. This work studies the memorization effect on incremental learning scenarios. Forgetting prevention and memorization seem similar. However, one should discuss their differences. We designed extensive experiments to evaluate the impact of memorization on continual learning. We clarified that learning examples with high memorization scores are forgotten faster than regular samples. Our findings also indicated that memorization is necessary to achieve the highest performance. However, at low memory regimes, forgetting regular samples is more important. We showed that the importance of a high-memorization score sample rises with an increase in the buffer size. We introduced a memorization proxy and employed it in the buffer policy problem to showcase how memorization could be used during incremental training. We demonstrated that including samples with a higher proxy memorization score is beneficial when the buffer size is large.

Jędrzej Kozal, Michał Woźniak

Purpose: We propose a novel method for continual learning based on the increasing depth of neural networks. This work explores whether extending neural network depth may be beneficial in a life-long learning setting. Methods: We propose a novel approach based on adding new layers on top of existing ones to enable the forward transfer of knowledge and adapting previously learned representations. We employ a method of determining the most similar tasks for selecting the best location in our network to add new nodes with trainable parameters. This approach allows for creating a tree-like model, where each node is a set of neural network parameters dedicated to a specific task. The Progressive Neural Network concept inspires the proposed method. Therefore, it benefits from dynamic changes in network structure. However, Progressive Neural Network allocates a lot of memory for the whole network structure during the learning process. The proposed method alleviates this by adding only part of a network for a new task and utilizing a subset of previously trained weights. At the same time, we may retain the benefit of PNN, such as no forgetting guaranteed by design, without needing a memory buffer. Results: Experiments on Split CIFAR and Split Tiny ImageNet show that the proposed algorithm is on par with other continual learning methods. In a more challenging setup with a single computer vision dataset as a separate task, our method outperforms Experience Replay. Conclusion: It is compatible with commonly used computer vision architectures and does not require a custom network structure. As an adaptation to changing data distribution is made by expanding the architecture, there is no need to utilize a rehearsal buffer. For this reason, our method could be used for sensitive applications where data privacy must be considered.

Michał Woźniak, Paweł Zyblewski, Paweł Ksieniewicz

One of the significant problems of streaming data classification is the occurrence of concept drift, consisting of the change of probabilistic characteristics of the classification task. This phenomenon destabilizes the performance of the classification model and seriously degrades its quality. An appropriate strategy counteracting this phenomenon is required to adapt the classifier to the changing probabilistic characteristics. One of the significant problems in implementing such a solution is the access to data labels. It is usually costly, so to minimize the expenses related to this process, learning strategies based on semi-supervised learning are proposed, e.g., employing active learning methods indicating which of the incoming objects are valuable to be labeled for improving the classifier's performance. This paper proposes a novel chunk-based method for non-stationary data streams based on classifier ensemble learning and an active learning strategy considering a limited budget that can be successfully applied to any data stream classification algorithm. The proposed method has been evaluated through computer experiments using both real and generated data streams. The results confirm the high quality of the proposed algorithm over state-of-the-art methods.

Jędrzej Kozal, Filip Guzy, Michał Woźniak

Modern analytical systems must be ready to process streaming data and correctly respond to data distribution changes. The phenomenon of changes in data distributions is called concept drift, and it may harm the quality of the used models. Additionally, the possibility of concept drift appearance causes that the used algorithms must be ready for the continuous adaptation of the model to the changing data distributions. This work focuses on non-stationary data stream classification, where a classifier ensemble is used. To keep the ensemble model up to date, the new base classifiers are trained on the incoming data blocks and added to the ensemble while, at the same time, outdated models are removed from the ensemble. One of the problems with this type of model is the fast reaction to changes in data distributions. We propose a new Chunk Adaptive Restoration framework that can be adapted to any block-based data stream classification algorithm. The proposed algorithm adjusts the data chunk size in the case of concept drift detection to minimize the impact of the change on the predictive performance of the used model. The conducted experimental research, backed up with the statistical tests, has proven that Chunk Adaptive Restoration significantly reduces the model's restoration time.

Michał Koziarski, Colin Bellinger, Michał Woźniak

Real-world classification domains, such as medicine, health and safety, and finance, often exhibit imbalanced class priors and have asynchronous misclassification costs. In such cases, the classification model must achieve a high recall without significantly impacting precision. Resampling the training data is the standard approach to improving classification performance on imbalanced binary data. However, the state-of-the-art methods ignore the local joint distribution of the data or correct it as a post-processing step. This can causes sub-optimal shifts in the training distribution, particularly when the target data distribution is complex. In this paper, we propose Radial-Based Combined Cleaning and Resampling (RB-CCR). RB-CCR utilizes the concept of class potential to refine the energy-based resampling approach of CCR. In particular, RB-CCR exploits the class potential to accurately locate sub-regions of the data-space for synthetic oversampling. The category sub-region for oversampling can be specified as an input parameter to meet domain-specific needs or be automatically selected via cross-validation. Our $5\times2$ cross-validated results on 57 benchmark binary datasets with 9 classifiers show that RB-CCR achieves a better precision-recall trade-off than CCR and generally out-performs the state-of-the-art resampling methods in terms of AUC and G-mean.

Joanna Grzyb, Jakub Klikowski, Michał Woźniak

The imbalanced data classification remains a vital problem. The key is to find such methods that classify both the minority and majority class correctly. The paper presents the classifier ensemble for classifying binary, non-stationary and imbalanced data streams where the Hellinger Distance is used to prune the ensemble. The paper includes an experimental evaluation of the method based on the conducted experiments. The first one checks the impact of the base classifier type on the quality of the classification. In the second experiment, the Hellinger Distance Weighted Ensemble (HDWE) method is compared to selected state-of-the-art methods using a statistical test with two base classifiers. The method was profoundly tested based on many imbalanced data streams and obtained results proved the HDWE method's usefulness.

Michał Koziarski, Michał Woźniak, Bartosz Krawczyk

The imbalanced data classification is one of the most crucial tasks facing modern data analysis. Especially when combined with other difficulty factors, such as the presence of noise, overlapping class distributions, and small disjuncts, data imbalance can significantly impact the classification performance. Furthermore, some of the data difficulty factors are known to affect the performance of the existing oversampling strategies, in particular SMOTE and its derivatives. This effect is especially pronounced in the multi-class setting, in which the mutual imbalance relationships between the classes complicate even further. Despite that, most of the contemporary research in the area of data imbalance focuses on the binary classification problems, while their more difficult multi-class counterparts are relatively unexplored. In this paper, we propose a novel oversampling technique, a Multi-Class Combined Cleaning and Resampling (MC-CCR) algorithm. The proposed method utilizes an energy-based approach to modeling the regions suitable for oversampling, less affected by small disjuncts and outliers than SMOTE. It combines it with a simultaneous cleaning operation, the aim of which is to reduce the effect of overlapping class distributions on the performance of the learning algorithms. Finally, by incorporating a dedicated strategy of handling the multi-class problems, MC-CCR is less affected by the loss of information about the inter-class relationships than the traditional multi-class decomposition strategies. Based on the results of experimental research carried out for many multi-class imbalanced benchmark datasets, the high robust of the proposed approach to noise was shown, as well as its high quality compared to the state-of-art methods.

José-Ramón Cano, Pedro Antonio Gutiérrez, Bartosz Krawczyk et al.

Currently, knowledge discovery in databases is an essential step to identify valid, novel and useful patterns for decision making. There are many real-world scenarios, such as bankruptcy prediction, option pricing or medical diagnosis, where the classification models to be learned need to fulfil restrictions of monotonicity (i.e. the target class label should not decrease when input attributes values increase). For instance, it is rational to assume that a higher debt ratio of a company should never result in a lower level of bankruptcy risk. Consequently, there is a growing interest from the data mining research community concerning monotonic predictive models. This paper aims to present an overview about the literature in the field, analyzing existing techniques and proposing a taxonomy of the algorithms based on the type of model generated. For each method, we review the quality metrics considered in the evaluation and the different data sets and monotonic problems used in the analysis. In this way, this paper serves as an overview of the research about monotonic classification in specialized literature and can be used as a functional guide of the field.