Urban growth, particularly in the Global South, poses significant challenges due to the emergence of informal settlements as well as planned areas. Existing urban growth models, including multi-agent systems, often rely on rigid spatial structures and strong hypothesis, which are poorly suited for both contexts. This paper introduces a novel method that combines procedural generation and genetic algorithms to model spatial dynamics in agent-based urban growth simulations. By defining two types of spatial influence functions and optimizing their parameters using a genetic algorithm (NSGA-II), our approach eliminates the need for empirical assumptions and improves realism by learning from spatial data. Experiments demonstrate the applicability of this method in generating realistic informal settlements and planned areas, validated through a combination of density and distance measures. This study contributes to the field by providing a more flexible and accurate model for simulating complex urban growth. • Introduction of a generic agent-based model for complex urban growth. • Definition of spatial influence functions to model environmental factors. • Introduction of a novel method combining procedural generation and genetic algorithms for urban growth simulation. • Validation of the model through a combination of spatial measures estimating density and distance differences. • Demonstration of the method’s applicability in generating realistic informal settlements and planned areas.

International audience

Programming language processing is a field of increasing interest, as more and more models become available, either to address specific tasks or to acquire general knowledge which can then be fine-tuned on downstream tasks. All these models are based on architectures that come from the field of natural language processing, most of them being built on the distributional hypothesis from linguistics. Although this transition from one field to another appears to have occurred naturally, it is not so obvious to claim that this hypothesis will be appropriate for extracting semantics from programs. In this paper, we investigate to which extent a distributional hypothesis can be applied to code embedding. To this end, we first formulate various hypotheses adapted to the specific information contained in programming languages. We then provide a framework to evaluate the effectiveness of these hypotheses through the quality of the resulting embedding spaces. This framework is based on the doc2vec model as a generic language model, as its implementation of the original distributional hypothesis is easy to understand and to adapt to any new ones. Among other tools, we propose a new evaluation method based on program analogies, which measures how well the models capture the underlying structure and meaning of the code. We apply the proposed framework to a set of (distributional) hypotheses and show that we can rule out certain hypotheses in favor of others. Specifically, our study indicates that instruction-based hypotheses capture less semantic information than token-based ones. Furthermore, we observe that distributional hypotheses on tokens are effective in both source code, execution traces, and abstract syntax trees. Additionally, we find that the semantics captured on programs by these three hypotheses are of comparable levels and natures.

International audience

Data mining methods extract knowledge from huge amounts of data. Recently with the explosion of mobile technologies, a new type of data appeared. The resulting databases can be described as spatiotemporal data in which spatial information (e.g., the location of an event) and temporal information (e .g., the date of the event) are included. In this article, we focus on spatiotemporal patterns extraction from this kind of databases. These patterns can be considered as sequences representing changes of events localized in areas and its near surrounding over time. Two algorithms are proposed to tackle this problem: the first one uses \emph{a priori} strategy and the second one is based on pattern-growth approach. We have applied our generic method on two different real datasets related to: 1) pollution of rivers in France; and 2) monitoring of dengue epidemics in New Caledonia. Additionally, experiments on synthetic data have been conducted to measure the performance of the proposed algorithms.

To support knowledge discovery from data, many pattern mining techniques have been proposed. One of the bottlenecks for their dissemination is the number of computed patterns that appear to be either trivial or uninteresting with respect to available knowledge. Integration of domain knowledge in constraint-based data mining is limited. Relevant patterns still miss because methods partly fail in assessing their subjective interestingness. However, in practice, we often have in the literature mathematical models defined by experts based on their domain knowledge. We propose here to exploit such models to derive constraints that can be used during the data mining phase to improve both pattern relevancy and computational efficiency. Even though the approach is generic, it is illustrated on pattern set discovery from real data for studying soil erosion.

Extraction of interesting colocations in geo-referenced data is one of the major tasks in spatial pattern mining. The goal is to find sets of spatial object-types with instances located in the same neighborhood. In this context, the main drawback is the visualization and interpretation of extracted patterns by domain experts. Indeed, common textual representation of colocations loses important spatial information such as the position, the orientation or the spatial distribution of the patterns. To overcome this problem, we propose a new clustering-based visualization technique deeply integrated in the colocation mining algorithm. This new simple, concise and intuitive cartographic visualization considers both spatial information and expert practices. This proposition has been integrated in a Geographic Information System and experimented on a real-world geological data set. Domain experts confirm the added-value of this visualization approach.

Interesting pattern mining is an important family of data mining problems with applications in many domains. In the last decade, many algorithms and benchmarks have been devised and many experimental studies have been performed for pattern mining problems. In this paper, we focus on the special class of pattern mining problems known to be ‘representable as sets’. This class of problem is important and have many applications such as in data mining, in databases, in artificial intelligence, or in software engineering.

The discovery of frequent patterns is a famous problem in data mining. While plenty of algorithms have been proposed during the last decade, only a few contributions have tried to understand the influence of datasets on the algorithms behavior. Being able to explain why certain algorithms are likely to perform very well or very poorly on some datasets is still an open question. In this setting, we describe a thorough experimental study of datasets with respect to frequent item sets. We study the distribution of frequent item sets with respect to item sets size together with the distribution of three concise representations: frequent closed, frequent free and frequent essential item sets. For each of them, we also study the distribution of their positive and negative borders whenever possible. From this analysis, we exhibit a new characterization of datasets and some invariants allowing to better predict the behavior of well known algorithms. The main perspective of this work is to devise adaptive algorithms with respect to dataset characteristics.