Hyperspectral Analysis of Columbia Spotted Frog Habitat

Abstract: Wildlife managers increasingly are using remotely sensed imagery to improve habitat delineations and sampling strategies. Advances in remote sensing technology, such as hyperspectral imagery, provide more information than previously was available with multispectral sensors. We evaluated accuracy of high-resolution hyperspectral image classifications to identify wetlands and wetland habitat features important for Columbia spotted frogs (Rana luteiventris) and compared the results to multispectral image classification and United States Geological Survey topographic maps. The study area spanned 3 lake basins in the Salmon River Mountains, Idaho, USA. Hyperspectral data were collected with an airborne sensor on 30 June 2002 and on 8 July 2006. A 12-year comprehensive ground survey of the study area for Columbia spotted frog reproduction served as validation for image classifications. Hyperspectral image classification accuracy of wetlands was high, with a producer's accuracy of 96% (44 wetlands) correctly classified with the 2002 data and 89% (41 wetlands) correctly classified with the 2006 data. We applied habitat-based rules to delineate breeding habitat from other wetlands, and successfully predicted 74% (14 wetlands) of known breeding wetlands for the Columbia spotted frog. Emergent sedge microhabitat classification showed promise for directly predicting Columbia spotted frog egg mass locations within a wetland by correctly identifying 72% (23 of 32) of known locations. Our study indicates hyperspectral imagery can be an effective tool for mapping spotted frog breeding habitat in the selected mountain basins. We conclude that this technique has potential for improving site selection for inventory and monitoring programs conducted across similar wetland habitat and can be a useful tool for delineating wildlife habitats.     

Weighting Class Importance in Agricultural Crop Classification from Remotely Sensed Data with an Artificial Neural Network

In some classifications the importance of classes varies and it is desirable to weight allocation to selected classes. This is common in classifications of remotely sensed imagery, especially as class occurrence can vary markedly. If, for instance, there is prior knowledge on the distribution of class occurrence this weighting can be achieved with widely used statistical classifiers by setting appropriate a priori probabilities of class membership. With an arificial neural network the incorporation of prior knowledge is more problematic. An approach to weight class allocation in an artificial neural network classifcation by replicating selected training patterns is discussed. In comparison against a discriminant analysis for the classification of synthetic aperture radar imagery the results showed that training pattern replication could be used to weight class allocation with an effect similar to that of incorporating a priori probabilities of class membership into the discriminant analysis and resulted in a significant increase in classification accuracy.     

Classification of Camellia (Theaceae) species using leaf architecture variations and pattern recognition techniques

Leaf characters have been successfully utilized to classify Camellia (Theaceae) species; however, leaf characters combined with supervised pattern recognition techniques have not been previously explored. We present results of using leaf morphological and venation characters of 93 species from five sections of genus Camellia to assess the effectiveness of several supervised pattern recognition techniques for classifications and compare their accuracy. Clustering approach, Learning Vector Quantization neural network (LVQ-ANN), Dynamic Architecture for Artificial Neural Networks (DAN2), and C-support vector machines (SVM) are used to discriminate 93 species from five sections of genus Camellia (11 in sect. Furfuracea, 16 in sect. Paracamellia, 12 in sect. Tuberculata, 34 in sect. Camellia, and 20 in sect. Theopsis). DAN2 and SVM show excellent classification results for genus Camellia with DAN2's accuracy of 97.92% and 91.11% for training and testing data sets respectively. The RBF-SVM results of 97.92% and 97.78% for training and testing offer the best classification accuracy. A hierarchical dendrogram based on leaf architecture data has confirmed the morphological classification of the five sections as previously proposed. The overall results suggest that leaf architecture-based data analysis using supervised pattern recognition techniques, especially DAN2 and SVM discrimination methods, is excellent for identification of Camellia species.     

The relevance of non-farmland habitats, uncropped areas and habitat diversity to the conservation of farmland birds

Use of non-farmland habitats by species generally perceived as 'farmland birds' is common, yet these habitats are not always considered in conservation strategies aimed at population recovery. At the national scale, many farmland species occur in landscapes not dominated by farmland. An analysis of distribution atlas data coupled with remotely sensed habitat data showed that for 16 out of 28 farmland species, less than half of the breeding range was associated with high cover of lowland farmland. However, with a few exceptions, populations breeding in non-farmland habitats are likely to depend on farmland at some time in the year. Within farmland landscapes, uncropped areas and patches of non-farmland habitat can provide nesting, foraging or roosting resources. Habitats that are scarce on farmland and that provide potential supplementary or complementary resources to those available within the productive areas of farmland include ruderal vegetation, rough grassland and scrub. Enhancing habitat diversity through provision of modest quantities of these habitats will benefit farmland birds. Complete knowledge of year-round habitat requirements and patterns of resource use at all scales is needed if robust national conservation plans are to be developed for farmland species. Similarly, interactions between the farmland and non-farmland sections of populations need to be determined.     

Differentiating geological fertility derived vegetation zones in Kruger National Park,South Africa,using Landsat and MODIS imagery

Spatial technologies present possibilities for producing frequently updated and accurate habitat maps, which are important in biodiversity conservation. Assemblages of vegetation are equivalent to habitats. This study examined the use of satellite imagery in vegetation differentiation in South Africa's Kruger National Park (KNP). A vegetation classification scheme based on dominant tree species but also related to the park's geology was tested, the geology generally consisting of high and low fertility lithology. Currently available multispectral satellite imagery is broadly either of high spatial but low temporal resolution or low spatial but high temporal resolution. Landsat TM/ETM+ and MODIS images were used to represent these broad categories. Rain season dates were selected as the period when discrimination between key habitats in KNP is most likely to be successful. Principal Component Analysis enhanced vegetated areas on the Landsat images, while NDVI vegetation enhancement was employed on the MODIS image. The images were classified into six field sampling derived classes depicting a vegetation density and phenology gradient, with high (about 89%) indicative classification accuracy. The results indicate that, using image processing procedures that enhance vegetation density, image classification can be used to map the park's vegetation at the high versus low geological fertility zone level, to accuracies above 80% on high spatial resolution imagery and slightly lower accuracy on lower spatial resolution imagery. Rainfall just prior to the image date influences herbaceous vegetation and, therefore, success at image scene vegetation mapping, while cloud cover limits image availability. Small scale habitat differentiation using multispectral satellite imagery for large protected savanna areas appears feasible, indicating the potential for use of remote sensing in savanna habitat monitoring. However, factors affecting successful habitat mapping need to be considered. Therefore, adoption of remote sensing in vegetation mapping and monitoring for large protected savanna areas merits consideration by conservation agencies.     

A novel method of protein secondary structure prediction with high segment overlap measure: support vector machine approach

We have introduced a new method of protein secondary structure prediction which is based on the theory of support vector machine (SVM). SVM represents a new approach to supervised pattern classification which has been successfully applied to a wide range of pattern recognition problems, including object recognition, speaker identification, gene function prediction with microarray expression profile, etc. In these cases, the performance of SVM either matches or is significantly better than that of traditional machine learning approaches, including neural networks.The first use of the SVM approach to predict protein secondary structure is described here. Unlike the previous studies, we first constructed several binary classifiers, then assembled a tertiary classifier for three secondary structure states (helix, sheet and coil) based on these binary classifiers. The SVM method achieved a good performance of segment overlap accuracy SOV=76.2 % through sevenfold cross validation on a database of 513 non-homologous protein chains with multiple sequence alignments, which out-performs existing methods. Meanwhile three-state overall per-residue accuracy Q(3) achieved 73.5 %, which is at least comparable to existing single prediction methods. Furthermore a useful "reliability index" for the predictions was developed. In addition, SVM has many attractive features, including effective avoidance of overfitting, the ability to handle large feature spaces, information condensing of the given data set, etc. The SVM method is conveniently applied to many other pattern classification tasks in biology.     

Remote sensing imagery in vegetation mapping: a review

Aims Mapping vegetation through remotely sensed images involves various considerations, processes and techniques. Increasing availability of remotely sensed images due to the rapid advancement of remote sensing technology expands the horizon of our choices of imagery sources. Various sources of imagery are known for their differences in spectral, spatial, radioactive and temporal characteristics and thus are suitable for different purposes of vegetation mapping. Generally, it needs to develop a vegetation classification at first for classifying and mapping vegetation cover from remote sensed images either at a community level or species level. Then, correlations of the vegetation types (communities or species) within this classification system with discernible spectral characteristics of remote sensed imagery have to be identified. These spectral classes of the imagery are finally translated into the vegetation types in the image interpretation process, which is also called image processing. This paper presents an overview of how to use remote sensing imagery to classify and map vegetation cover.Methods Specifically, this paper focuses on the comparisons of popular remote sensing sensors, commonly adopted image processing methods and prevailing classification accuracy assessments.Important findings The basic concepts, available imagery sources and classification techniques of remote sensing imagery related to vegetation mapping were introduced, analyzed and compared. The advantages and limitations of using remote sensing imagery for vegetation cover mapping were provided to iterate the importance of thorough understanding of the related concepts and careful design of the technical procedures, which can be utilized to study vegetation cover from remote sensed images.     

Real-Time Subject-Independent Pattern Classification of Overt and Covert Movements from fNIRS Signals

Recently, studies have reported the use of Near Infrared Spectroscopy (NIRS) for developing Brain–Computer Interface (BCI) by applying online pattern classification of brain states from subject-specific fNIRS signals. The purpose of the present study was to develop and test a real-time method for subject-specific and subject-independent classification of multi-channel fNIRS signals using support-vector machines (SVM), so as to determine its feasibility as an online neurofeedback system. Towards this goal, we used left versus right hand movement execution and movement imagery as study paradigms in a series of experiments. In the first two experiments, activations in the motor cortex during movement execution and movement imagery were used to develop subject-dependent models that obtained high classification accuracies thereby indicating the robustness of our classification method. In the third experiment, a generalized classifier-model was developed from the first two experimental data, which was then applied for subject-independent neurofeedback training. Application of this method in new participants showed mean classification accuracy of 63% for movement imagery tasks and 80% for movement execution tasks. These results, and their corresponding offline analysis reported in this study demonstrate that SVM based real-time subject-independent classification of fNIRS signals is feasible. This method has important applications in the field of hemodynamic BCIs, and neuro-rehabilitation where patients can be trained to learn spatio-temporal patterns of healthy brain activity.     

Mapping potential habitats of threatened plant species in a moist tall grassland using hyperspectral imagery

We examined the capability of hyperspectral imagery to map habitat types of under-storey plants in a moist tall grassland dominated by Phragmites australis and Miscanthus sacchariflorus, using hyperspectral remotely-sensed shoot densities of the two grasses. Our procedure (1) grouped the species using multivariate analysis and discriminated habitat types (species groups) based on P. australis and M. sacchariflorus shoot densities, (2) used estimated shoot densities from hyperspectral data to draw a habitat type map, and (3) analyzed the association of threatened species with habitat types. Our identification of four habitat types, using cluster analysis of the vegetation survey coverage data, was based on P. australis and M. sacchariflorus shoot density ratios and had an overall accuracy of 77.1% (kappa coefficient = 0.71). Linear regression models based on hyperspectral imagery band data had good accuracy in estimating P. australis and M. sacchariflorus shoot densities (adjusted R ² = 0.686 and 0.708, respectively). These results enabled us to map under-storey plant habitat types to an approximate prediction accuracy of 0.537. Among the eight threatened species we examined, four exhibited a significantly biased distribution among habitat types, indicating species-specific habitat use. These results suggest that this procedure can provide useful information on the status of potential habitats of threatened species.     

Patterns of bird diversity and habitat use in mixed vineyard-matorral landscapes of Central Chile

The Mediterranean climate region of central Chile is rich in biodiversity and contains highly productive agricultural lands, which creates challenges for the preservation of natural habitats and native biodiversity. Ecological data and studies for the region are also limited, making informed conservation in agricultural landscapes difficult. The increasing availability of remotely sensed data provide opportunities to relate species occurrences to measures of landscape heterogeneity even when field measures of habitat structure are lacking. When working with such remotely sensed data, it’s important to select appropriate measures of heterogeneity, including common metrics of landscape composition as well as frequently overlooked shape metrics. In this contribution we combine bird surveys with multispectral satellite imagery to develop boosted regression tree models of avian species richness, and of habitat use for 15 species across a mixed vineyard-matorral landscape in central Chile. We found a range of associations between individual species and land cover types, with the majority of species occurring most frequently in remnant habitats and ecotones rather than the interiors of large vineyard blocks. Models identified both metrics of landscape composition and patch shape as being important predictors of species occurrence, suggesting that shape metrics can complement more commonly used metrics of landscape composition. Vineyards that include corridors or islands of remnant habitat among vine blocks may increase the amount of area available to many species, although some species may still require large tracts of intact natural habitat to persist.     

Classification of pasture habitats by Hungarian herders in a steppe landscape (Hungary)

ABSTRACT: BACKGROUND: Landscape ethnoecology focuses on the ecological features of the landscape, how the landscape is perceived, and used by people who live in it. Though studying folk classifications of species has a long history, the comparative study of habitat classifications is just beginning. I studied the habitat classification of herders in a Hungarian steppe, and compared it to classifications of botanists and laymen. METHODS: For a quantitative analysis the picture sort method was used. Twenty-three pictures of 7-11 habitat types were sorted by 25 herders. 'Density' of pictures along the habitat gradient of the Hortobagy salt steppe was set as equal as possible, but pictures differed in their dominant species, wetness, season, etc. Before sorts, herders were asked to describe pictures to assure proper recognition of habitats. RESULTS: Herders classified the images into three main (and 6 smaller) groups: (1) fertile habitats at the higher parts of the habitat gradient (partos, lit. on the shore); (2) saline habitats (szik, lit. salt or saline place), and (3) meadows and marshes (lapos, lit. flooded) at the lower end of the habitat gradient. Sharpness of delimitation changed along the gradient. Saline habitats were the most isolated from the rest. Botanists identified 6 groups. Laymen grouped habitats in a less coherent way. As opposed to my expectations, botanical classification was not more structured than that done by herders. I expected and found high correspondence between the classifications by herders, botanists and laymen. All tended to recognize similar main groups: wetlands, "good grass" and dry/saline habitats. Two main factors could have been responsible for similar classifications: salient features correlated (e.g. salinity recognizable by herders and botanists but not by laymen correlated with the density of grasslands or height of vegetation recognizable also for laymen), or the same salient features were used as a basis for sorting (wetness, and abiotic stress). CONCLUSIONS: Despite all the difficulties of studying habitat classifications (more implicit, more variable knowledge than knowledge on species), conducting landscape ethnoecological research will inevitably reveal a deeper human understanding of biological organization at a supraspecific level, where natural discontinuities are less sharp than at the species or population level.     

Prediction of cell penetrating peptides by support vector machines

Cell penetrating peptides (CPPs) are those peptides that can transverse cell membranes to enter cells. Once inside the cell, different CPPs can localize to different cellular components and perform different roles. Some generate pore-forming complexes resulting in the destruction of cells while others localize to various organelles. Use of machine learning methods to predict potential new CPPs will enable more rapid screening for applications such as drug delivery. We have investigated the influence of the composition of training datasets on the ability to classify peptides as cell penetrating using support vector machines (SVMs). We identified 111 known CPPs and 34 known non-penetrating peptides from the literature and commercial vendors and used several approaches to build training data sets for the classifiers. Features were calculated from the datasets using a set of basic biochemical properties combined with features from the literature determined to be relevant in the prediction of CPPs. Our results using different training datasets confirm the importance of a balanced training set with approximately equal number of positive and negative examples. The SVM based classifiers have greater classification accuracy than previously reported methods for the prediction of CPPs, and because they use primary biochemical properties of the peptides as features, these classifiers provide insight into the properties needed for cell-penetration. To confirm our SVM classifications, a subset of peptides classified as either penetrating or non-penetrating was selected for synthesis and experimental validation. Of the synthesized peptides predicted to be CPPs, 100% of these peptides were shown to be penetrating.     

Object-oriented methods for habitat mapping at multiple scales – Case studies from Northern Germany and Wye Downs, UK

This paper presents an application of object-oriented techniques for habitat classification based on remotely sensed images and ancillary data. The study reports the results of habitat mapping at multiple scales using Earth Observation (EO) data at various spatial resolutions and multi temporal acquisition dates. We investigate the role of object texture and context in classification as well as the value of integrating knowledge from ancillary data sources. Habitat maps were produced at regional and local scales in two case studies; Schleswig-Holstein, Germany and Wye Downs, United Kingdom. At the regional scale, the main task was the development of a consistent object-oriented classification scheme that is transferable to satellite images for other years. This is demonstrated for a time series of Landsat TM/ETM+ scenes. At the local scale, investigations focus on the development of appropriate object-oriented rule networks for the detailed mapping of habitats, e.g. dry grasslands and wetlands using very high resolution satellite and airborne scanner images. The results are evaluated using statistical accuracy assessment and visual comparison with traditional field-based habitat maps. Whereas the application of traditional pixel-based classification result in a pixelised (salt and pepper) representation of land cover, the object-based classification technique result in solid habitat objects allowing easy integration into a vector-GIS for further analysis. The level of detail obtained at the local scale is comparable to that achieved by visual interpretation of aerial photographs or field-based mapping and also retains spatially explicit, fine scale information such as scrub encroachment or ecotone patterns within habitats.     

Modelling avian biodiversity using raw,unclassified satellite imagery

Applications of remote sensing for biodiversity conservation typically rely on image classifications that do not capture variability within coarse land cover classes. Here, we compare two measures derived from unclassified remotely sensed data, a measure of habitat heterogeneity and a measure of habitat composition, for explaining bird species richness and the spatial distribution of 10 species in a semi-arid landscape of New Mexico. We surveyed bird abundance from 1996 to 1998 at 42 plots located in the McGregor Range of Fort Bliss Army Reserve. Normalized Difference Vegetation Index values of two May 1997 Landsat scenes were the basis for among-pixel habitat heterogeneity (image texture), and we used the raw imagery to decompose each pixel into different habitat components (spectral mixture analysis). We used model averaging to relate measures of avian biodiversity to measures of image texture and spectral mixture analysis fractions. Measures of habitat heterogeneity, particularly angular second moment and standard deviation, provide higher explanatory power for bird species richness and the abundance of most species than measures of habitat composition. Using image texture, alone or in combination with other classified imagery-based approaches, for monitoring statuses and trends in biological diversity can greatly improve conservation efforts and habitat management.     

How does dune morphology shape coastal EC habitats occurrence? A remote sensing approach using airborne LiDAR on the Mediterranean coast

We examined the relationship between coastal habitats (sensu European Union Habitats Directive) and local dune morphology along a Mediterranean coastal dune system by integrating field collected vegetation data and remotely sensed imagery. Specifically, we described the morphological profile of each EC habitat based on the morphological variables that are most likely to affect their occurrence, including elevation, slope, curvature, northness, eastness and sea distance. In addition, we assessed the role and strength of each morphological variable in determining the occurrence of EC habitats.We used 394 random vegetation plots representative of six EC habitats (Habitat 1210: “Annual vegetation of drift lines”; Habitat 2110: “Embryonic shifting dunes”; Habitat 2120: “Shifting dunes along the shoreline with Ammophila arenaria”; Habitat 2210 and 2230: “Crucianellion maritimae fixed beach dunes” and “Malcolmietalia dune grasslands”; Habitat 2250: “Coastal dunes with Juniperus spp.”; Habitat 2260: “Cisto-Lavanduletalia dune sclerophyllous scrubs”) found along the Tyrrhenian coast of central Italy. We derived each morphological variable from a DTM (Digital Terrain Model) obtained from 2-m resolution LiDAR (Light Detection And Range) images. The mean value of each variable was calculated at different spatial scales using buffer areas of increasing radius (2 m, 4 m, 8 m) around each vegetation plot. Mean morphological values for each EC habitat were compared using Kruskal-Wallis rank test. The role and strength of the relationship between habitat type and the morphological variables were assessed using Generalized Linear Models.EC habitats occur differentially across dune morphology, and the role and strength of each morphological variable define habitat specificity. Dune elevation and sea distance were determined to be the key factors in shaping EC habitat occurrence along this section of the Mediterranean coast. Identification of the close relationship between habitat type and morphological variables deriving from airborne LiDAR imagery points to the high potential of such remote sensing tool for analyzing and monitoring the integrity of coastal dune ecosystems. As airborne LiDAR enables the rapid collection of extremely accurate topographic data over large areas, it also offers useful information for the management of these threatened and fragile ecosystems.     

Mapping and monitoring of coastal wetlands of Çukurova Delta in the Eastern Mediterranean region

The aim of this research was to link vegetation characteristics, such as spatial and temporal distribution, and environmental variables, with land cover information derived from remotely sensed satellite images of the Eastern Mediterranean coastal wetlands of Turkey. The research method was based on (i) recording land cover characteristics by means of a vegetation indicator, and (ii) classifying and mapping coastal wetlands utilizing a Landsat Thematic Mapper (TM) image of Çukurova Deltas in Turkey. Vegetation characteristics of various habitats, such as sand dunes, salt marshes, salty plains and afforestation areas, were identified by field surveys. A Landsat TM image of 4 July 1993 was pre-processed and then classified using the Maximum Likelihood (ML) algorithm and Artificial Neural Networks (ANN). As a result of this supervised classification, the land cover types were classified with a largest accuracy of 90.2% by ANN. The classified satellite sensor imagery was linked to vegetation and bird census data, which were available through literature in a Geographical Information System (GIS) environment to determine the spatial distribution of plant and bird biodiversity in this coastal wetland. The resulting data provide an important baseline for further investigations such as monitoring, change detections and designing conservation policies in this coastal ecosystem.     

Predicting top predator habitats in the Southwest Indian Ocean

Top predators need to develop optimal strategies of resources and habitats utilization in order to optimize their foraging success. At the individual scale, a predator has to maximize his intake of food while minimizing his cost of foraging to optimize his energetic gain. At the ecosystem scale, we hypothesized that foraging strategies of predators also respond to their general energetic constraints. Predators with energetically costly lifestyles may be constrained to select high quality habitats whereas more phlegmatic predators may occupy both low and high quality habitats. The objectives of this study were 1) to investigate predator responses to heterogeneity in habitat quality with reference to their energetic strategies and 2) to evaluate their responses to contemporaneous versus averaged habitat quality. We collected cetacean and seabird data from an aerial survey in the Southwest Indian Ocean, a region characterized by heterogeneous oceanographic conditions. We classified cetaceans and seabirds into energetic guilds and described their habitats using remotely sensed covariates at contemporaneous and time‐averaged resolutions and static covariates. We used generalized additive models to predict their habitats at the regional scale. Strategies of habitat utilization appeared in accordance with predators energetic constraints. Cetaceans responded to the heterogeneity in habitat quality, with higher densities predicted in more productive areas. However, the costly Delphininae appeared to be more dependent on habitat quality (showing a 1‐to‐13 ratio between the lowest and highest density sectors) than the more phlegmatic sperm and beaked whales (showing only a 1‐to‐3 ratio). For seabirds, predictions primarily reflected colony locations, although the colony effect was stronger for costly seabirds. Moreover, our results suggest that predators may respond better to persistent oceanographic features. To provide a third dimension to habitat quality, cetacean strategies of utilization of the vertical habitat could be related to the distribution of micronekton in the water column.     

Remote sensing variables as predictors of habitat suitability of the viscacha rat (<Emphasis Type="Italic">Octomys mimax</Emphasis>), a rock-dwelling mammal living in a desert environment

Identifying high-quality habitats across large areas is a central goal in biodiversity conservation. Remotely sensed data provide the opportunity to study different habitat characteristics (e.g., landscape topography, soil, vegetation cover, climatic factors) that are difficult to identify at high spatial and temporal resolution on the basis of field studies. Our goal was to evaluate the applicability of remotely sensed information as a potential tool for modeling habitat suitability of the viscacha rat (Octomys mimax), a rock-dwelling species that lives in a desert ecosystem. We fitted models considering raw indices (i.e., green indices, Brightness Index (BI) and temperature) and their derived texture measures on locations used by and available for the viscacha rat. The habitat preferences identified in our models are consistent with results of field studies of landscape use by the viscacha rat. Rocky habitats were well differentiated by the second-order contrast of BI, instead of BI only, making an important contribution to the global model by capturing the heterogeneity of the substratum. Furthermore, rocky habitats are able to maintain more vegetation than much of the surrounding desert; hence, their availability might be estimated using SATVI (Soil Adjusted Total Vegetation Index) and its derived texture measures: second-order contrast and entropy. This is the first study that evaluates the usefulness of remotely sensed data for predicting and mapping habitat suitability for a small-bodied rock dwelling species in a desert environment. Our results may contribute to conservation efforts focused on these habitat specialist species by using good predictors of habitat quality.     

Decoding semi-constrained brain activity from FMRI using support vector machines and gaussian processes

Predicting a particular cognitive state from a specific pattern of fMRI voxel values is still a methodological challenge. Decoding brain activity is usually performed in highly controlled experimental paradigms characterized by a series of distinct states induced by a temporally constrained experimental design. In more realistic conditions, the number, sequence and duration of mental states are unpredictably generated by the individual, resulting in complex and imbalanced fMRI data sets. This study tests the classification of brain activity, acquired on 16 volunteers using fMRI, during mental imagery, a condition in which the number and duration of mental events were not externally imposed but self-generated. To deal with these issues, two classification techniques were considered (Support Vector Machines, SVM, and Gaussian Processes, GP), as well as different feature extraction methods (General Linear Model, GLM and SVM). These techniques were combined in order to identify the procedures leading to the highest accuracy measures. Our results showed that 12 data sets out of 16 could be significantly modeled by either SVM or GP. Model accuracies tended to be related to the degree of imbalance between classes and to task performance of the volunteers. We also conclude that the GP technique tends to be more robust than SVM to model unbalanced data sets.