Piotr Szymański

Piotr Gramacki, Kacper Leśniara, Kamil Raczycki et al.

Spatial Representations for Artificial Intelligence (srai) is a Python library for working with geospatial data. The library can download geospatial data, split a given area into micro-regions using multiple algorithms and train an embedding model using various architectures. It includes baseline models as well as more complex methods from published works. Those capabilities make it possible to use srai in a complete pipeline for geospatial task solving. The proposed library is the first step to standardize the geospatial AI domain toolset. It is fully open-source and published under Apache 2.0 licence.

Łukasz Augustyniak, Kamil Tagowski, Albert Sawczyn et al.

The availability of compute and data to train larger and larger language models increases the demand for robust methods of benchmarking the true progress of LM training. Recent years witnessed significant progress in standardized benchmarking for English. Benchmarks such as GLUE, SuperGLUE, or KILT have become de facto standard tools to compare large language models. Following the trend to replicate GLUE for other languages, the KLEJ benchmark has been released for Polish. In this paper, we evaluate the progress in benchmarking for low-resourced languages. We note that only a handful of languages have such comprehensive benchmarks. We also note the gap in the number of tasks being evaluated by benchmarks for resource-rich English/Chinese and the rest of the world. In this paper, we introduce LEPISZCZE (the Polish word for glew, the Middle English predecessor of glue), a new, comprehensive benchmark for Polish NLP with a large variety of tasks and high-quality operationalization of the benchmark. We design LEPISZCZE with flexibility in mind. Including new models, datasets, and tasks is as simple as possible while still offering data versioning and model tracking. In the first run of the benchmark, we test 13 experiments (task and dataset pairs) based on the five most recent LMs for Polish. We use five datasets from the Polish benchmark and add eight novel datasets. As the paper's main contribution, apart from LEPISZCZE, we provide insights and experiences learned while creating the benchmark for Polish as the blueprint to design similar benchmarks for other low-resourced languages.

Piotr Kawa, Marcin Plata, Michał Czuba et al.

With a recent influx of voice generation methods, the threat introduced by audio DeepFake (DF) is ever-increasing. Several different detection methods have been presented as a countermeasure. Many methods are based on so-called front-ends, which, by transforming the raw audio, emphasize features crucial for assessing the genuineness of the audio sample. Our contribution contains investigating the influence of the state-of-the-art Whisper automatic speech recognition model as a DF detection front-end. We compare various combinations of Whisper and well-established front-ends by training 3 detection models (LCNN, SpecRNet, and MesoNet) on a widely used ASVspoof 2021 DF dataset and later evaluating them on the DF In-The-Wild dataset. We show that using Whisper-based features improves the detection for each model and outperforms recent results on the In-The-Wild dataset by reducing Equal Error Rate by 21%.

Kacper Leśniara, Piotr Szymański

Recent years brought advancements in using neural networks for representation learning of various language or visual phenomena. New methods freed data scientists from hand-crafting features for common tasks. Similarly, problems that require considering the spatial variable can benefit from pretrained map region representations instead of manually creating feature tables that one needs to prepare to solve a task. However, very few methods for map area representation exist, especially with respect to road network characteristics. In this paper, we propose a method for generating microregions' embeddings with respect to their road infrastructure characteristics. We base our representations on OpenStreetMap road networks in a selection of cities and use the H3 spatial index to allow reproducible and scalable representation learning. We obtained vector representations that detect how similar map hexagons are in the road networks they contain. Additionally, we observe that embeddings yield a latent space with meaningful arithmetic operations. Finally, clustering methods allowed us to draft a high-level typology of obtained representations. We are confident that this contribution will aid data scientists working on infrastructure-related prediction tasks with spatial variables.

Kamil Raczycki, Marcin Szymański, Yahor Yeliseyenka et al.

Public transport agencies use social media as an essential tool for communicating mobility incidents to passengers. However, while the short term, day-to-day information about transport phenomena is usually posted in social media with low latency, its availability is short term as the content is rarely made an aggregated form. Social media communication of transport phenomena usually lacks GIS annotations as most social media platforms do not allow attaching non-POI GPS coordinates to posts. As a result, the analysis of transport phenomena information is minimal. We collected three years of social media posts of a polish public transport company with user comments. Through exploration, we infer a six-class transport information typology. We successfully build an information type classifier for social media posts, detect stop names in posts, and relate them to GPS coordinates, obtaining a spatial understanding of long-term aggregated phenomena. We show that our approach enables citizen science and use it to analyze the impact of three years of infrastructure incidents on passenger mobility, and the sentiment and reaction scale towards each of the events. All these results are achieved for Polish, an under-resourced language when it comes to spatial language understanding, especially in social media contexts. To improve the situation, we released two of our annotated data sets: social media posts with incident type labels and matched stop names and social media comments with the annotated sentiment. We also opensource the experimental codebase.

Julia Moska, Oleksii Furman, Kacper Kozaczko et al.

GeoAI is evolving rapidly, fueled by diverse geospatial datasets like traffic patterns, environmental data, and crowdsourced OpenStreetMap (OSM) information. While sophisticated AI models are being developed, existing benchmarks are often concentrated on single tasks and restricted to a single modality. As such, progress in GeoAI is limited by the lack of a standardized, multi-task, modality-agnostic benchmark for their systematic evaluation. This paper introduces a novel benchmark designed to assess the performance, accuracy, and efficiency of geospatial embedders. Our benchmark is modality-agnostic and comprises 7 distinct datasets from diverse cities across three continents, ensuring generalizability and mitigating demographic biases. It allows for the evaluation of GeoAI embedders on various phenomena that exhibit underlying geographic processes. Furthermore, we establish a simple and intuitive task-oriented model baselines, providing a crucial reference point for comparing more complex solutions.

Bruno Martins, Piotr Szymański, Piotr Gramacki

The emergence of Large Language Models (LLMs) has transformed information access, with current LLMs also powering deep research systems that can generate comprehensive report-style answers, through planned iterative search, retrieval, and reasoning. Still, current deep research systems lack the geo-temporal capabilities that are essential for answering context-rich questions involving geographic and/or temporal constraints, frequently occurring in domains like public health, environmental science, or socio-economic analysis. This paper reports our vision towards next generation systems, identifying important technical, infrastructural, and evaluative challenges in integrating geo-temporal reasoning into deep research pipelines. We argue for augmenting retrieval and synthesis processes with the ability to handle geo-temporal constraints, supported by open and reproducible infrastructures and rigorous evaluation protocols. Our vision outlines a path towards more advanced and geo-temporally aware deep research systems, of potential impact to the future of AI-driven information access.

Kamil Raczycki, Piotr Szymański

Bicycle-sharing systems (BSS) have become a daily reality for many citizens of larger, wealthier cities in developed regions. However, planning the layout of bicycle-sharing stations usually requires expensive data gathering, surveying travel behavior and trip modelling followed by station layout optimization. Many smaller cities and towns, especially in developing areas, may have difficulty financing such projects. Planning a BSS also takes a considerable amount of time. Yet as the pandemic has shown us, municipalities will face the need to adapt rapidly to mobility shifts, which include citizens leaving public transport for bicycles. Laying out a bike sharing system quickly will become critical in addressing the increase in bike demand. This paper addresses the problem of cost and time in BSS layout design and proposes a new solution to streamline and facilitate the process of such planning by using spatial embedding methods. Based only on publicly available data from OpenStreetMap, and station layouts from 34 cities in Europe, a method has been developed to divide cities into micro-regions using the Uber H3 discrete global grid system and to indicate regions where it is worth placing a station based on existing systems in different cities using transfer learning. The result of the work is a mechanism to support planners in their decision making when planning a station layout with a choice of reference cities.

Szymon Woźniak, Piotr Szymański

Representation learning of spatial and geographic data is a rapidly developing field which allows for similarity detection between areas and high-quality inference using deep neural networks. Past approaches however concentrated on embedding raster imagery (maps, street or satellite photos), mobility data or road networks. In this paper we propose the first approach to learning vector representations of OpenStreetMap regions with respect to urban functions and land-use in a micro-region grid. We identify a subset of OSM tags related to major characteristics of land-use, building and urban region functions, types of water, green or other natural areas. Through manual verification of tagging quality, we selected 36 cities were for training region representations. Uber's H3 index was used to divide the cities into hexagons, and OSM tags were aggregated for each hexagon. We propose the hex2vec method based on the Skip-gram model with negative sampling. The resulting vector representations showcase semantic structures of the map characteristics, similar to ones found in vector-based language models. We also present insights from region similarity detection in six Polish cities and propose a region typology obtained through agglomerative clustering.

Piotr Gramacki, Szymon Woźniak, Piotr Szymański

We selected 48 European cities and gathered their public transport timetables in the GTFS format. We utilized Uber's H3 spatial index to divide each city into hexagonal micro-regions. Based on the timetables data we created certain features describing the quantity and variety of public transport availability in each region. Next, we trained an auto-associative deep neural network to embed each of the regions. Having such prepared representations, we then used a hierarchical clustering approach to identify similar regions. To do so, we utilized an agglomerative clustering algorithm with a euclidean distance between regions and Ward's method to minimize in-cluster variance. Finally, we analyzed the obtained clusters at different levels to identify some number of clusters that qualitatively describe public transport availability. We showed that our typology matches the characteristics of analyzed cities and allows succesful searching for areas with similar public transport schedule characteristics.

Piotr Szymański, Piotr Żelasko, Mikolaj Morzy et al.

Natural language processing of conversational speech requires the availability of high-quality transcripts. In this paper, we express our skepticism towards the recent reports of very low Word Error Rates (WERs) achieved by modern Automatic Speech Recognition (ASR) systems on benchmark datasets. We outline several problems with popular benchmarks and compare three state-of-the-art commercial ASR systems on an internal dataset of real-life spontaneous human conversations and HUB'05 public benchmark. We show that WERs are significantly higher than the best reported results. We formulate a set of guidelines which may aid in the creation of real-life, multi-domain datasets with high quality annotations for training and testing of robust ASR systems.

Piotr Szymański, Kyle Gorman

Recent work raises concerns about the use of standard splits to compare natural language processing models. We propose a Bayesian statistical model comparison technique which uses k-fold cross-validation across multiple data sets to estimate the likelihood that one model will outperform the other, or that the two will produce practically equivalent results. We use this technique to rank six English part-of-speech taggers across two data sets and three evaluation metrics.

Jan Mizgajski, Adrian Szymczak, Robert Głowski et al.

Avaya Conversational Intelligence(ACI) is an end-to-end, cloud-based solution for real-time Spoken Language Understanding for call centers. It combines large vocabulary, real-time speech recognition, transcript refinement, and entity and intent recognition in order to convert live audio into a rich, actionable stream of structured events. These events can be further leveraged with a business rules engine, thus serving as a foundation for real-time supervision and assistance applications. After the ingestion, calls are enriched with unsupervised keyword extraction, abstractive summarization, and business-defined attributes, enabling offline use cases, such as business intelligence, topic mining, full-text search, quality assurance, and agent training. ACI comes with a pretrained, configurable library of hundreds of intents and a robust intent training environment that allows for efficient, cost-effective creation and customization of customer-specific intents.

Piotr Żelasko, Jan Mizgajski, Mikołaj Morzy et al.

In this paper, we present a method for correcting automatic speech recognition (ASR) errors using a finite state transducer (FST) intent recognition framework. Intent recognition is a powerful technique for dialog flow management in turn-oriented, human-machine dialogs. This technique can also be very useful in the context of human-human dialogs, though it serves a different purpose of key insight extraction from conversations. We argue that currently available intent recognition techniques are not applicable to human-human dialogs due to the complex structure of turn-taking and various disfluencies encountered in spontaneous conversations, exacerbated by speech recognition errors and scarcity of domain-specific labeled data. Without efficient key insight extraction techniques, raw human-human dialog transcripts remain significantly unexploited. Our contribution consists of a novel FST for intent indexing and an algorithm for fuzzy intent search over the lattice - a compact graph encoding of ASR's hypotheses. We also develop a pruning strategy to constrain the fuzziness of the FST index search. Extracted intents represent linguistic domain knowledge and help us improve (rescore) the original transcript. We compare our method with a baseline, which uses only the most likely transcript hypothesis (best path), and find an increase in the total number of recognized intents by 25%.

Piotr Szymański, Tomasz Kajdanowicz, Nitesh Chawla

Multi-label classification aims to classify instances with discrete non-exclusive labels. Most approaches on multi-label classification focus on effective adaptation or transformation of existing binary and multi-class learning approaches but fail in modelling the joint probability of labels or do not preserve generalization abilities for unseen label combinations. To address these issues we propose a new multi-label classification scheme, LNEMLC - Label Network Embedding for Multi-Label Classification, that embeds the label network and uses it to extend input space in learning and inference of any base multi-label classifier. The approach allows capturing of labels' joint probability at low computational complexity providing results comparable to the best methods reported in the literature. We demonstrate how the method reveals statistically significant improvements over the simple kNN baseline classifier. We also provide hints for selecting the robust configuration that works satisfactorily across data domains.

Piotr Żelasko, Piotr Szymański, Jan Mizgajski et al.

An ASR system usually does not predict any punctuation or capitalization. Lack of punctuation causes problems in result presentation and confuses both the human reader andoff-the-shelf natural language processing algorithms. To overcome these limitations, we train two variants of Deep Neural Network (DNN) sequence labelling models - a Bidirectional Long Short-Term Memory (BLSTM) and a Convolutional Neural Network (CNN), to predict the punctuation. The models are trained on the Fisher corpus which includes punctuation annotation. In our experiments, we combine time-aligned and punctuated Fisher corpus transcripts using a sequence alignment algorithm. The neural networks are trained on Common Web Crawl GloVe embedding of the words in Fisher transcripts aligned with conversation side indicators and word time infomation. The CNNs yield a better precision and BLSTMs tend to have better recall. While BLSTMs make fewer mistakes overall, the punctuation predicted by the CNN is more accurate - especially in the case of question marks. Our results constitute significant evidence that the distribution of words in time, as well as pre-trained embeddings, can be useful in the punctuation prediction task.

Piotr Szymański, Tomasz Kajdanowicz

In the recent years, we have witnessed the development of multi-label classification methods which utilize the structure of the label space in a divide and conquer approach to improve classification performance and allow large data sets to be classified efficiently. Yet most of the available data sets have been provided in train/test splits that did not account for maintaining a distribution of higher-order relationships between labels among splits or folds. We present a new approach to stratifying multi-label data for classification purposes based on the iterative stratification approach proposed by Sechidis et. al. in an ECML PKDD 2011 paper. Our method extends the iterative approach to take into account second-order relationships between labels. Obtained results are evaluated using statistical properties of obtained strata as presented by Sechidis. We also propose new statistical measures relevant to second-order quality: label pairs distribution, the percentage of label pairs without positive evidence in folds and label pair - fold pairs that have no positive evidence for the label pair. We verify the impact of new methods on classification performance of Binary Relevance, Label Powerset and a fast greedy community detection based label space partitioning classifier. Random Forests serve as base classifiers. We check the variation of the number of communities obtained per fold, and the stability of their modularity score. Second-Order Iterative Stratification is compared to standard k-fold, label set, and iterative stratification. The proposed approach lowers the variance of classification quality, improves label pair oriented measures and example distribution while maintaining a competitive quality in label-oriented measures. We also witness an increase in stability of network characteristics.

Piotr Szymański, Tomasz Kajdanowicz

We study the performance of data-driven, a priori and random approaches to label space partitioning for multi-label classification with a Gaussian Naive Bayes classifier. Experiments were performed on 12 benchmark data sets and evaluated on 5 established measures of classification quality: micro and macro averaged F1 score, Subset Accuracy and Hamming loss. Data-driven methods are significantly better than an average run of the random baseline. In case of F1 scores and Subset Accuracy - data driven approaches were more likely to perform better than random approaches than otherwise in the worst case. There always exists a method that performs better than a priori methods in the worst case. The advantage of data-driven methods against a priori methods with a weak classifier is lesser than when tree classifiers are used.

Piotr Szymański, Tomasz Kajdanowicz

scikit-multilearn is a Python library for performing multi-label classification. The library is compatible with the scikit/scipy ecosystem and uses sparse matrices for all internal operations. It provides native Python implementations of popular multi-label classification methods alongside a novel framework for label space partitioning and division. It includes modern algorithm adaptation methods, network-based label space division approaches, which extracts label dependency information and multi-label embedding classifiers. It provides python wrapped access to the extensive multi-label method stack from Java libraries and makes it possible to extend deep learning single-label methods for multi-label tasks. The library allows multi-label stratification and data set management. The implementation is more efficient in problem transformation than other established libraries, has good test coverage and follows PEP8. Source code and documentation can be downloaded from http://scikit.ml and also via pip. The library follows BSD licensing scheme.

Piotr Szymański, Tomasz Kajdanowicz, Kristian Kersting

We propose using five data-driven community detection approaches from social networks to partition the label space for the task of multi-label classification as an alternative to random partitioning into equal subsets as performed by RAkELd: modularity-maximizing fastgreedy and leading eigenvector, infomap, walktrap and label propagation algorithms. We construct a label co-occurence graph (both weighted an unweighted versions) based on training data and perform community detection to partition the label set. We include Binary Relevance and Label Powerset classification methods for comparison. We use gini-index based Decision Trees as the base classifier. We compare educated approaches to label space divisions against random baselines on 12 benchmark data sets over five evaluation measures. We show that in almost all cases seven educated guess approaches are more likely to outperform RAkELd than otherwise in all measures, but Hamming Loss. We show that fastgreedy and walktrap community detection methods on weighted label co-occurence graphs are 85-92% more likely to yield better F1 scores than random partitioning. Infomap on the unweighted label co-occurence graphs is on average 90% of the times better than random paritioning in terms of Subset Accuracy and 89% when it comes to Jaccard similarity. Weighted fastgreedy is better on average than RAkELd when it comes to Hamming Loss.