Cluster-based active learning for compact image classification

In this paper, we consider active sampling to label pixels grouped with hierarchical clustering. The objective of the method is to match the data relationships discovered by the clustering algorithm with the user's desired class semantics. The first is represented as a complete tree to be pruned and the second is iteratively provided by the user. The active learning algorithm proposed searches the pruning of the tree that best matches the labels of the sampled points. By choosing the part of the tree to sample from according to current pruning's uncertainty, sampling is focused on most uncertain clusters. This way, large clusters for which the class membership is already fixed are no longer queried and sampling is focused on division of clusters showing mixed labels. The model is tested on a VHR image in a multiclass classification setting. The method clearly outperforms random sampling in a transductive setting, but cannot generalize to unseen data, since it aims at optimizing the classification of a given cluster structure.

The Effect of Adding CO2 to Hypoxic Inspired Gas on Cerebral Blood Flow Velocity and Breathing during Incremental Exercise

Hypoxia increases the ventilatory response to exercise, which leads to hyperventilation-induced hypocapnia and subsequent reduction in cerebral blood flow (CBF). We studied the effects of adding CO2 to a hypoxic inspired gas on CBF during heavy exercise in an altitude naïve population. We hypothesized that augmented inspired CO2 and hypoxia would exert synergistic effects on increasing CBF during exercise, which would improve exercise capacity compared to hypocapnic hypoxia. We also examined the responsiveness of CO2 and O2 chemoreception on the regulation ventilation (E) during incremental exercise. We measured middle cerebral artery velocity (MCAv; index of CBF), E, end-tidal PCO2, respiratory compensation threshold (RC) and ventilatory response to exercise (E slope) in ten healthy men during incremental cycling to exhaustion in normoxia and hypoxia (FIO2 = 0.10) with and without augmenting the fraction of inspired CO2 (FICO2). During exercise in normoxia, augmenting FICO2 elevated MCAv throughout exercise and lowered both RC onset andE slope below RC (P<0.05). In hypoxia, MCAv and E slope below RC during exercise were elevated, while the onset of RC occurred at lower exercise intensity (P<0.05). Augmenting FICO2 in hypoxia increased E at RC (P<0.05) but no difference was observed in RC onset, MCAv, or E slope below RC (P>0.05). The E slope above RC was unchanged with either hypoxia or augmented FICO2 (P>0.05). We found augmenting FICO2 increased CBF during sub-maximal exercise in normoxia, but not in hypoxia, indicating that the 'normal' cerebrovascular response to hypercapnia is blunted during exercise in hypoxia, possibly due to an exhaustion of cerebral vasodilatory reserve. This finding may explain the lack of improvement of exercise capacity in hypoxia with augmented CO2. Our data further indicate that, during exercise below RC, chemoreception is responsive, while above RC the ventilatory response to CO2 is blunted.

Learning through Computer Model Improvisations

It has been convincingly argued that computer simulation modeling differs from traditional science. If we understand simulation modeling as a new way of doing science, the manner in which scientists learn about the world through models must also be considered differently. This article examines how researchers learn about environmental processes through computer simulation modeling. Suggesting a conceptual framework anchored in a performative philosophical approach, we examine two modeling projects undertaken by research teams in England, both aiming to inform flood risk management. One of the modeling teams operated in the research wing of a consultancy firm, the other were university scientists taking part in an interdisciplinary project experimenting with public engagement. We found that in the first context the use of standardized software was critical to the process of improvisation, the obstacles emerging in the process concerned data and were resolved through exploiting affordances for generating, organizing, and combining scientific information in new ways. In the second context, an environmental competency group, obstacles were related to the computer program and affordances emerged in the combination of experience-based knowledge with the scientists' skill enabling a reconfiguration of the mathematical structure of the model, allowing the group to learn about local flooding.

Transferring control demands across incidental learning tasks - Stronger sequence usage in serial reaction task after shortcut option in letter string checking

After incidentally learning about a hidden regularity, participants can either continue to solve the task as instructed or, alternatively, apply a shortcut. Past research suggests that the amount of conflict implied by adopting a shortcut seems to bias the decision for vs. against continuing instruction-coherent task processing. We explored whether this decision might transfer from one incidental learning task to the next. Theories that conceptualize strategy change in incidental learning as a learning-plus-decision phenomenon suggest that high demands to adhere to instruction-coherent task processing in Task 1 will impede shortcut usage in Task 2, whereas low control demands will foster it. We sequentially applied two established incidental learning tasks differing in stimuli, responses and hidden regularity (the alphabet verification task followed by the serial reaction task, SRT). While some participants experienced a complete redundancy in the task material of the alphabet verification task (low demands to adhere to instructions), for others the redundancy was only partial. Thus, shortcut application would have led to errors (high demands to follow instructions). The low control demand condition showed the strongest usage of the fixed and repeating sequence of responses in the SRT. The transfer results are in line with the learning-plus-decision view of strategy change in incidental learning, rather than with resource theories of self-control.

A model for the evolution of reinforcement learning in fluctuating games

Many species are able to learn to associate behaviours with rewards as this gives fitness advantages in changing environments. Social interactions between population members may, however, require more cognitive abilities than simple trial-and-error learning, in particular the capacity to make accurate hypotheses about the material payoff consequences of alternative action combinations. It is unclear in this context whether natural selection necessarily favours individuals to use information about payoffs associated with nontried actions (hypothetical payoffs), as opposed to simple reinforcement of realized payoff. Here, we develop an evolutionary model in which individuals are genetically determined to use either trial-and-error learning or learning based on hypothetical reinforcements, and ask what is the evolutionarily stable learning rule under pairwise symmetric two-action stochastic repeated games played over the individual's lifetime. We analyse through stochastic approximation theory and simulations the learning dynamics on the behavioural timescale, and derive conditions where trial-and-error learning outcompetes hypothetical reinforcement learning on the evolutionary timescale. This occurs in particular under repeated cooperative interactions with the same partner. By contrast, we find that hypothetical reinforcement learners tend to be favoured under random interactions, but stable polymorphisms can also obtain where trial-and-error learners are maintained at a low frequency. We conclude that specific game structures can select for trial-and-error learning even in the absence of costs of cognition, which illustrates that cost-free increased cognition can be counterselected under social interactions.

Learning-induced gene expression in the heads of two Nasonia species that differ in long-term memory formation.

BACKGROUND: Cellular processes underlying memory formation are evolutionary conserved, but natural variation in memory dynamics between animal species or populations is common. The genetic basis of this fascinating phenomenon is poorly understood. Closely related species of Nasonia parasitic wasps differ in long-term memory (LTM) formation: N. vitripennis will form transcription-dependent LTM after a single conditioning trial, whereas the closely-related species N. giraulti will not. Genes that were differentially expressed (DE) after conditioning in N. vitripennis, but not in N. giraulti, were identified as candidate genes that may regulate LTM formation. RESULTS: RNA was collected from heads of both species before and immediately, 4 or 24 hours after conditioning, with 3 replicates per time point. It was sequenced strand-specifically, which allows distinguishing sense from antisense transcripts and improves the quality of expression analyses. We determined conditioning-induced DE compared to naïve controls for both species. These expression patterns were then analysed with GO enrichment analyses for each species and time point, which demonstrated an enrichment of signalling-related genes immediately after conditioning in N. vitripennis only. Analyses of known LTM genes and genes with an opposing expression pattern between the two species revealed additional candidate genes for the difference in LTM formation. These include genes from various signalling cascades, including several members of the Ras and PI3 kinase signalling pathways, and glutamate receptors. Interestingly, several other known LTM genes were exclusively differentially expressed in N. giraulti, which may indicate an LTM-inhibitory mechanism. Among the DE transcripts were also antisense transcripts. Furthermore, antisense transcripts aligning to a number of known memory genes were detected, which may have a role in regulating these genes. CONCLUSION: This study is the first to describe and compare expression patterns of both protein-coding and antisense transcripts, at different time points after conditioning, of two closely related animal species that differ in LTM formation. Several candidate genes that may regulate differences in LTM have been identified. This transcriptome analysis is a valuable resource for future in-depth studies to elucidate the role of candidate genes and antisense transcription in natural variation in LTM formation.

Information, learning, and risk in financial markets

Cette thèse comprend trois essais qui abordent l'information le processus d'ap-prentissage ainsi que le risque dans les marchés finances. Elle se concentre d'abord sur les implications à l'équilibre de l'hétérogénéité des agents à travers un processus d'apprentissage comprtemental et de mise à jour de l'information. De plus, elle examine les effets du partage des risques dans un reseau entreprise-fournisseur. Le premier chapitre étudie les effets du biais de disponibili sur l'évaluation des actifs. Ce biais décrit le fait que les agents surestiment l'importance de l'information acquise via l'expérience personnelle. L'hétérogénéité restante des différentes perceptions individuelles amène à une volonté d'échanges. Conformé¬ment aux données empiriques, les jeunes agents échangent plus mais en même temps souffrent d'une performance inférieure. Le deuxième chapitre se penche sur l'impact qu'ont les différences de modelisation entre les agents sur leurs percevons individuelles du processus de prix, dans le contexte des projections de modèles. Les agents sujets à un biais de projection pensent être représentatifs et interprètent les opinions des autres agents comme du bruit. Les agents, avec des modèles plus persistants, perçoivent que les prix réagissent de façon excessive lors des périodes de turbulence. Le troisième chapitre analyse l'impact du partage des risques dans la relation entreprise-fournisseur sur la décision optimale de financement de l'entreprise. Il étudie l'impact sur l'optimisation de la structure du capital ainsi que sur le coût du capital. Les résultats indiquent en particulier qu'un fournisseur avec un effet de levier faible est utile pour le financement d'un nouveau projet d'investissement. Pour des projets très rentables et des fournisseurs à faible effet de levier, le coût des capitaux propres de l'entreprise peut diminuer.

From bird to sparrow: Learning-induced modulations in fine-grained semantic discrimination.

Recognition of environmental sounds is believed to proceed through discrimination steps from broad to more narrow categories. Very little is known about the neural processes that underlie fine-grained discrimination within narrow categories or about their plasticity in relation to newly acquired expertise. We investigated how the cortical representation of birdsongs is modulated by brief training to recognize individual species. During a 60-minute session, participants learned to recognize a set of birdsongs; they improved significantly their performance for trained (T) but not control species (C), which were counterbalanced across participants. Auditory evoked potentials (AEPs) were recorded during pre- and post-training sessions. Pre vs. post changes in AEPs were significantly different between T and C i) at 206-232ms post stimulus onset within a cluster on the anterior part of the left superior temporal gyrus; ii) at 246-291ms in the left middle frontal gyrus; and iii) 512-545ms in the left middle temporal gyrus as well as bilaterally in the cingulate cortex. All effects were driven by weaker activity for T than C species. Thus, expertise in discriminating T species modulated early stages of semantic processing, during and immediately after the time window that sustains the discrimination between human vs. animal vocalizations. Moreover, the training-induced plasticity is reflected by the sharpening of a left lateralized semantic network, including the anterior part of the temporal convexity and the frontal cortex. Training to identify birdsongs influenced, however, also the processing of C species, but at a much later stage. Correct discrimination of untrained sounds seems to require an additional step which results from lower-level features analysis such as apperception. We therefore suggest that the access to objects within an auditory semantic category is different and depends on subject's level of expertise. More specifically, correct intra-categorical auditory discrimination for untrained items follows the temporal hierarchy and transpires in a late stage of semantic processing. On the other hand, correct categorization of individually trained stimuli occurs earlier, during a period contemporaneous with human vs. animal vocalization discrimination, and involves a parallel semantic pathway requiring expertise.