arXiv subjects

Krzysztof Krawiec

Publications and source records attributed to Krzysztof Krawiec.

18 recordsLinked to original sources

arXiv2026

Structure over Pixels: Learning Variable-Length Visual Programs

Discrete visual tokenizers map images to ordered sequences of tokens, providing a natural representation for structural scene descriptions. Most use a fixed sequence length, while adaptive methods often require post-hoc search or choose among a small set of rates that control the length. We propose STROP, a discrete tokenizer that learns both a visual program and its image-dependent active length. A length head is trained with a four-phase curriculum using local rate-distortion probes against frozen DINOv3 features, then predicts the active prefix in a single forward pass. At a matched rate of about $250$ nominal bits per crop, the adaptive model improves segmentation over a separately trained fixed-length baseline on four benchmarks (by $1.6$-$3.1$ mIoU), and it also beats a fixed $K{=}32$ baseline that uses more bits. STROP programs also yield higher segmentation mIoU than FlexTok, One-D-Piece, and ALIT at similar or higher rates, under the same readout architecture and training protocol. STROP therefore learns useful per-image sequence lengths without post-hoc search or a predefined set of compression rates.

cs.CV↗

arXiv2026

Improving Generative Adversarial Networks with Self-Distillation

In modern GANs, maintaining an Exponential Moving Average (EMA) of the generator's weights is a standard practice, as such an averaged model consistently outperforms the actively trained generator. However, the EMA generator is used for final deployment only and does not influence the training process. To address this missed opportunity, we introduce Self-Distilled GAN (SD-GAN) that employs the EMA generator as a teacher to guide the active generator (student) via perceptual loss. We prove the local asymptotic stability of SD-GAN in the Dirac-GAN setting and show that it dampens the parasitic cycling behavior that plagues the conventional GANs. Empirical evaluations across established architectures and datasets demonstrate that SD-GAN improves the final image quality on several metrics (FID and random-FID in particular), stabilizes the optimization trajectory and provides additional learning guidance that is not trivially correlated with the conventional adversarial loss. It also proves effective for fine-tuning pretrained GAN models.

cs.CV↗

arXiv2026

Physics-Informed Spectral Modeling for Hyperspectral Imaging

We present PhISM, a physics-informed deep learning architecture that learns without supervision to explicitly disentangle hyperspectral observations and model them with continuous basis functions. PhISM outperforms prior methods on several classification and regression benchmarks, requires limited labeled data, and provides additional insights thanks to interpretable latent representation.

cs.LG↗

arXiv2026

Autoassociative Learning of Structural Representations for Modeling and Classification in Medical Imaging

Deep learning architectures based on convolutional neural networks tend to rely on continuous, smooth features. While this characteristics provides significant robustness and proves useful in many real-world tasks, it is strikingly incompatible with the physical characteristic of the world, which, at the scale in which humans operate, comprises crisp objects, typically representing well-defined categories. This study proposes a class of neurosymbolic systems that learn by reconstructing images in terms of visual primitives and are thus forced to form high-level, structural explanations of them. When applied to the task of diagnosing abnormalities in histological imaging, the method proved superior to a conventional deep learning architecture in terms of classification accuracy, while being more transparent.

cs.CV↗

arXiv2025

Modeling Retinal Ganglion Cells with Neural Differential Equations

This work explores Liquid Time-Constant Networks (LTCs) and Closed-form Continuous-time Networks (CfCs) for modeling retinal ganglion cell activity in tiger salamanders across three datasets. Compared to a convolutional baseline and an LSTM, both architectures achieved lower MAE, faster convergence, smaller model sizes, and favorable query times, though with slightly lower Pearson correlation. Their efficiency and adaptability make them well suited for scenarios with limited data and frequent retraining, such as edge deployments in vision prosthetics.

cs.CV↗

arXiv2025

Generative Learning of Differentiable Object Models for Compositional Interpretation of Complex Scenes

This study builds on the architecture of the Disentangler of Visual Priors (DVP), a type of autoencoder that learns to interpret scenes by decomposing the perceived objects into independent visual aspects of shape, size, orientation, and color appearance. These aspects are expressed as latent parameters which control a differentiable renderer that performs image reconstruction, so that the model can be trained end-to-end with gradient using reconstruction loss. In this study, we extend the original DVP so that it can handle multiple objects in a scene. We also exploit the interpretability of its latent by using the decoder to sample additional training examples and devising alternative training modes that rely on loss functions defined not only in the image space, but also in the latent space. This significantly facilitates training, which is otherwise challenging due to the presence of extensive plateaus in the image-space reconstruction loss. To examine the performance of this approach, we propose a new benchmark featuring multiple 2D objects, which subsumes the previously proposed Multi-dSprites dataset while being more parameterizable. We compare the DVP extended in these ways with two baselines (MONet and LIVE) and demonstrate its superiority in terms of reconstruction quality and capacity to decompose overlapping objects. We also analyze the gradients induced by the considered loss functions, explain how they impact the efficacy of training, and discuss the limitations of differentiable rendering in autoencoders and the ways in which they can be addressed.

cs.CV↗

arXiv2025

Solving rescheduling problems in heterogeneous urban railway networks using hybrid quantum-classical approach

We address the applicability of a hybrid quantum-classical heuristics for practical railway rescheduling management problems. We build an integer linear programming model and solve it with D-Wave's quantum-classical hybrid solver (CQM) as well as with CPLEX, for comparison. The proposed approach is demonstrated on a real-life heterogeneous urban network in Poland, including both single- and multi-track segments. All the requirements posed by the operator of the network are included in the model. The computational results demonstrate the readiness for application and the benefits of quantum-classical hybrid solvers in a realistic railway scenario: they yield acceptable solutions on time, which is a critical requirement in a rescheduling situation. In particular, CQM as a probabilistic heuristic solver provides a number of feasible, close-to-optimal solutions the dispatcher can choose from.

quant-ph↗

arXiv2025

Learning Semantics-aware Search Operators for Genetic Programming

Fitness landscapes in test-based program synthesis are known to be extremely rugged, with even minimal modifications of programs often leading to fundamental changes in their behavior and, consequently, fitness values. Relying on fitness as the only guidance in iterative search algorithms like genetic programming is thus unnecessarily limiting, especially when combined with purely syntactic search operators that are agnostic about their impact on program behavior. In this study, we propose a semantics-aware search operator that steers the search towards candidate programs that are valuable not only actually (high fitness) but also only potentially, i.e. are likely to be turned into high-quality solutions even if their current fitness is low. The key component of the method is a graph neural network that learns to model the interactions between program instructions and processed data, and produces a saliency map over graph nodes that represents possible search decisions. When applied to a suite of symbolic regression benchmarks, the proposed method outperforms conventional tree-based genetic programming and the ablated variant of the method.

cs.LG↗

arXiv2024

Guiding Genetic Programming with Graph Neural Networks

In evolutionary computation, it is commonly assumed that a search algorithm acquires knowledge about a problem instance by sampling solutions from the search space and evaluating them with a fitness function. This is necessarily inefficient because fitness reveals very little about solutions -- yet they contain more information that can be potentially exploited. To address this observation in genetic programming, we propose EvoNUDGE, which uses a graph neural network to elicit additional knowledge from symbolic regression problems. The network is queried on the problem before an evolutionary run to produce a library of subprograms, which is subsequently used to seed the initial population and bias the actions of search operators. In an extensive experiment on a large number of problem instances, EvoNUDGE is shown to significantly outperform multiple baselines, including the conventional tree-based genetic programming and the purely neural variant of the method.

cs.NE↗

arXiv2024

Learning to Solve Abstract Reasoning Problems with Neurosymbolic Program Synthesis and Task Generation

The ability to think abstractly and reason by analogy is a prerequisite to rapidly adapt to new conditions, tackle newly encountered problems by decomposing them, and synthesize knowledge to solve problems comprehensively. We present TransCoder, a method for solving abstract problems based on neural program synthesis, and conduct a comprehensive analysis of decisions made by the generative module of the proposed architecture. At the core of TransCoder is a typed domain-specific language, designed to facilitate feature engineering and abstract reasoning. In training, we use the programs that failed to solve tasks to generate new tasks and gather them in a synthetic dataset. As each synthetic task created in this way has a known associated program (solution), the model is trained on them in supervised mode. Solutions are represented in a transparent programmatic form, which can be inspected and verified. We demonstrate TransCoder's performance using the Abstract Reasoning Corpus dataset, for which our framework generates tens of thousands of synthetic problems with corresponding solutions and facilitates systematic progress in learning.

cs.AI↗

arXiv2024

Disentangling Visual Priors: Unsupervised Learning of Scene Interpretations with Compositional Autoencoder

Contemporary deep learning architectures lack principled means for capturing and handling fundamental visual concepts, like objects, shapes, geometric transforms, and other higher-level structures. We propose a neurosymbolic architecture that uses a domain-specific language to capture selected priors of image formation, including object shape, appearance, categorization, and geometric transforms. We express template programs in that language and learn their parameterization with features extracted from the scene by a convolutional neural network. When executed, the parameterized program produces geometric primitives which are rendered and assessed for correspondence with the scene content and trained via auto-association with gradient. We confront our approach with a baseline method on a synthetic benchmark and demonstrate its capacity to disentangle selected aspects of the image formation process, learn from small data, correct inference in the presence of noise, and out-of-sample generalization.

cs.CV↗

arXiv2024

Learning Abstract Visual Reasoning via Task Decomposition: A Case Study in Raven Progressive Matrices

Learning to perform abstract reasoning often requires decomposing the task in question into intermediate subgoals that are not specified upfront, but need to be autonomously devised by the learner. In Raven Progressive Matrices (RPM), the task is to choose one of the available answers given a context, where both the context and answers are composite images featuring multiple objects in various spatial arrangements. As this high-level goal is the only guidance available, learning to solve RPMs is challenging. In this study, we propose a deep learning architecture based on the transformer blueprint which, rather than directly making the above choice, addresses the subgoal of predicting the visual properties of individual objects and their arrangements. The multidimensional predictions obtained in this way are then directly juxtaposed to choose the answer. We consider a few ways in which the model parses the visual input into tokens and several regimes of masking parts of the input in self-supervised training. In experimental assessment, the models not only outperform state-of-the-art methods but also provide interesting insights and partial explanations about the inference. The design of the method also makes it immune to biases that are known to be present in some RPM benchmarks.

cs.AI↗

arXiv2022

Quadratic and Higher-Order Unconstrained Binary Optimization of Railway Rescheduling for Quantum Computing

As consequences of disruptions in railway traffic affect passenger experience/satisfaction, appropriate rerouting and/or rescheduling is necessary. These problems are known to be NP-hard, given the numerous restrictions of traffic nature. With the recent advances in quantum technologies, quantum annealing has become an alternative method to solve such optimization problems. To use quantum annealing, the problem needs to be encoded in QUBO (quadratic unconstrained binary optimization) or HOBO (higher-order binary optimization) formulation that can be recast as a QUBO. This paper introduces QUBO and HOBO representations for rescheduling problems of railway traffic management; the latter is a new approach up to our knowledge. This new approach takes into account not only the single-track lines but also the double- and multi-track lines, as well as stations composed of tracks and switches. We consider the conditions of minimal headway between trains, minimal stay on stations, track occupation, and rolling stock circulation. Furthermore, a hybrid quantum-classical procedure is presented that includes rerouting. We demonstrate the proof of concept implementation on the D-Wave Quantum Processing Unit and D-Wave hybrid solver.

cs.ET↗

arXiv2021

Quantum annealing in the NISQ era: railway conflict management

We are in the Noisy Intermediate-Scale Quantum (NISQ) devices' era, in which quantum hardware has become available for application in real-world problems. However, demonstrating the usefulness of such NISQ devices are still rare. In this work, we consider a practical railway dispatching problem: delay and conflict management on single-track railway lines. We examine the issue of train dispatching consequences caused by the arrival of an already delayed train to the network segment being considered. This problem is computationally hard and needs to be solved almost in real-time. We introduce a quadratic unconstrained binary optimization (QUBO) model of this problem, compatible with the emerging quantum annealing technology. The model's instances can be executed on present-day quantum annealers. As a proof-of-concept, we solve selected real-life problems from the Polish railway network using D-Wave quantum annealers. As a reference, we also provide solutions calculated with classical methods, including those relevant to the community (linear integer programming) and a sophisticated algorithm based on tensor networks for solving Ising instances. Our preliminary results illustrate the degree of difficulty of real-life railway instances for the current quantum annealing technology. Moreover, our analysis shows that the new generation of quantum annealers (the advantage system) perform much worse on those instances than its predecessor.

quant-ph↗

arXiv2021

Quantum computing approach to railway dispatching and conflict management optimization on single-track railway lines

In this work, we consider a practical railway dispatching problem: delay and conflict management on a single-track railway line. We examine the issue of train dispatching consequences caused by the arrival of an already delayed train to the segment being considered. This problem is computationally hard and often needs to be solved timely. Here, we introduce a quadratic unconstrained binary optimization (QUBO) model of the problem in question, compatible with the emerging quantum annealing technology. The model's instances can be executed on present-day quantum annealers. As a proof-of-concept, we solve selected real-life problems from the Polish railway network using D-Wave quantum annealers. As a reference, we also provide solutions calculated with classical methods, including those relevant to the community (linear integer programming) and a sophisticated algorithm based on tensor networks for solving QUBO problems.

cs.ET↗

arXiv2019

Ain't Nobody Got Time For Coding: Structure-Aware Program Synthesis From Natural Language

Program synthesis from natural language (NL) is practical for humans and, once technically feasible, would significantly facilitate software development and revolutionize end-user programming. We present SAPS, an end-to-end neural network capable of mapping relatively complex, multi-sentence NL specifications to snippets of executable code. The proposed architecture relies exclusively on neural components, and is trained on abstract syntax trees, combined with a pretrained word embedding and a bi-directional multi-layer LSTM for processing of word sequences. The decoder features a doubly-recurrent LSTM, for which we propose novel signal propagation schemes and soft attention mechanism. When applied to a large dataset of problems proposed in a previous study, SAPS performs on par with or better than the method proposed there, producing correct programs in over 92% of cases. In contrast to other methods, it does not require post-processing of the resulting programs, and uses a fixed-dimensional latent representation as the only interface between the NL analyzer and the source code generator.

cs.LG↗

arXiv2017

Learning to Play Othello with Deep Neural Networks

Achieving superhuman playing level by AlphaGo corroborated the capabilities of convolutional neural architectures (CNNs) for capturing complex spatial patterns. This result was to a great extent due to several analogies between Go board states and 2D images CNNs have been designed for, in particular translational invariance and a relatively large board. In this paper, we verify whether CNN-based move predictors prove effective for Othello, a game with significantly different characteristics, including a much smaller board size and complete lack of translational invariance. We compare several CNN architectures and board encodings, augment them with state-of-the-art extensions, train on an extensive database of experts' moves, and examine them with respect to move prediction accuracy and playing strength. The empirical evaluation confirms high capabilities of neural move predictors and suggests a strong correlation between prediction accuracy and playing strength. The best CNNs not only surpass all other 1-ply Othello players proposed to date but defeat (2-ply) Edax, the best open-source Othello player.

cs.AI↗

arXiv2016

Distance Metric Ensemble Learning and the Andrews-Curtis Conjecture

Motivated by the search for a counterexample to the Poincaré conjecture in three and four dimensions, the Andrews-Curtis conjecture was proposed in 1965. It is now generally suspected that the Andrews-Curtis conjecture is false, but small potential counterexamples are not so numerous, and previous work has attempted to eliminate some via combinatorial search. Progress has however been limited, with the most successful approach (breadth-first-search using secondary storage) being neither scalable nor heuristically-informed. A previous empirical analysis of problem structure examined several heuristic measures of search progress and determined that none of them provided any useful guidance for search. In this article, we induce new quality measures directly from the problem structure and combine them to produce a more effective search driver via ensemble machine learning. By this means, we eliminate 19 potential counterexamples, the status of which had been unknown for some years.

cs.AI↗