Dataset Information

Development of swarm behavior in artificial learning agents that adapt to different foraging environments.

ABSTRACT: Collective behavior, and swarm formation in particular, has been studied from several perspectives within a large variety of fields, ranging from biology to physics. In this work, we apply Projective Simulation to model each individual as an artificial learning agent that interacts with its neighbors and surroundings in order to make decisions and learn from them. Within a reinforcement learning framework, we discuss one-dimensional learning scenarios where agents need to get to food resources to be rewarded. We observe how different types of collective motion emerge depending on the distance the agents need to travel to reach the resources. For instance, strongly aligned swarms emerge when the food source is placed far away from the region where agents are situated initially. In addition, we study the properties of the individual trajectories that occur within the different types of emergent collective dynamics. Agents trained to find distant resources exhibit individual trajectories that are in most cases best fit by composite correlated random walks with features that resemble Lévy walks. This composite motion emerges from the collective behavior developed under the specific foraging selection pressures. On the other hand, agents trained to reach nearby resources predominantly exhibit Brownian trajectories.

SUBMITTER: Lopez-Incera A

PROVIDER: S-EPMC7748156 | biostudies-literature | 2020

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Development of swarm behavior in artificial learning agents that adapt to different foraging environments.

López-Incera Andrea A Ried Katja K Müller Thomas T Briegel Hans J HJ

PloS one 20201218 12

Collective behavior, and swarm formation in particular, has been studied from several perspectives within a large variety of fields, ranging from biology to physics. In this work, we apply Projective Simulation to model each individual as an artificial learning agent that interacts with its neighbors and surroundings in order to make decisions and learn from them. Within a reinforcement learning framework, we discuss one-dimensional learning scenarios where agents need to get to food resources t ...[more]

PMID: 33338066

Similar Datasets

Project description:Snake venoms have evolved in a few families of Caenophidae, and their toxins have been fine-tuned over the evolutive process into highly effective biochemical weapons with a particular role as a trophic adaptation. However, there are many outstanding questions on how venom contributes to the success of venomous species and their adaptation to different environments. Here we analyze the venoms from sympatric and generalist specimens of B. hyoprora, B. taeniatus, B. b. smaragdinus, B. brazili, and B. atrox collected in the wild of Alto Juruá region at the Amazon Forest aiming to understand whether the venom composition could be a driver or influence the arboreal habitats of B. taeniatus and B. b. smaragdinus, or the successful dispersion of B. atrox across the Amazon Forest. Venom composition and the primary sequences of toxin isoforms were characterized by venom gland transcriptomics followed by proteomics. The venom composition of the five species conserved the same protein families present in venoms of Bothropoid snakes but with remarkable differences in the chromatographic profiles and in the relative amount of each toxin group: CTLs were the most abundant in the venoms of B. taeniatus (31.6%), B. b. smaragdinus (33.0%), and B. atrox (32.0%) and SVMPs and PLA2s dominated in venoms of B. hyoprora (23.7% and 20.2%) and B. brazili (20.2% and 20.4%). Some peculiarities were also observed: B. hyoprora venom was the least complex, conserving important isoforms shared with B. atrox venom; the presence of one cluster of SVSPs exclusive to B. brazili venom isoforms; and the presence of the K49-PLA2 homologs only in B. b. smaragdinus and particularly in B. brazili venoms. The SVMP, SVSP, and PLA2 enzyme activities were consistent with the abundance and type of isoforms present in the venoms. B. hyoprora, B. taeniatus, B. b. smaragdinus, and B. atrox venoms presented low lethality to small rodents, while B. brazili venom was the most lethal. Concluding, differences were evidenced between the venoms of B. taeniatus and B. b. smaragdinus arboreal species, while B. b. smaragdinus shared a venom with a similar composition to B. atrox venom, mostly terrestrial. Considering the role of venom toxicity in species' success for adaptation in different areas, B. atrox is the most adapted in the whole Amazonian territory and presented the least lethal venom, while the venom of B. brazili was the most lethal despite its small distribution in the same area. Therefore, our data suggest that the selection of venom composition may not influence the success or behavior of snake species. In this way, other biotic or abiotic factors influence their foraging status or their dispersal in different ecological niches.

Project description:Pollinators and the pollination services they provide are critical for seed set and self-sustainability of most flowering plants. Despite this, pollinators are rarely assessed in restored plant communities, where their services are largely assumed to re-establish. Bird-pollinator richness, foraging, and interaction behavior were compared between natural and restored Banksia woodland sites in Western Australia to assess their re-establishment in restored sites. These parameters were measured for natural communities of varying size and degree of fragmentation, and restored plant communities of high and low complexity for three years, in the summer and winter flowering of Banksia attenuata and B. menziesii, respectively. Bird visitor communities varied in composition, richness, foraging movement distances, and aggression among sites. Bird richness and abundance were lowest in fragmented remnants. Differences in the composition were associated with the size and degree of fragmentation in natural sites, but this did not differ between seasons. Restored sites and their adjacent natural sites had similar species composition, suggesting proximity supports pollinator re-establishment. Pollinator foraging movements were influenced by the territorial behavior of different species. Using a network analysis approach, we found foraging behavior varied, with more frequent aggressive chases observed in restored sites, resulting in more movements out of the survey areas, than observed in natural sites. Aggressors were larger-bodied Western Wattlebirds (Anthochaera chrysoptera) and New Holland Honeyeaters (Phylidonyris novaehollandiae) that dominated nectar resources, particularly in winter. Restored sites had re-established pollination services, albeit with clear differences, as the degree of variability in the composition and behavior of bird pollinators for Banksias in the natural sites created a broad completion target against which restored sites were assessed. The abundance, diversity, and behavior of pollinator services to remnant and restored Banksia woodland sites were impacted by the size and degree of fragmentation, which in turn influenced bird-pollinator composition, and were further influenced by seasonal changes between summer and winter. Consideration of the spatial and temporal landscape context of restored sites, along with plant community diversity, is needed to ensure the maintenance of the effective movement of pollinators between natural remnant woodlands and restored sites.

Dataset Information

Development of swarm behavior in artificial learning agents that adapt to different foraging environments.

Publications

Development of swarm behavior in artificial learning agents that adapt to different foraging environments.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets