Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “representation learning”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 847 records · Page 47Linked to original sources
PubMed2004

Muscular and postural synergies of the human hand.

Because humans have limited ability to independently control the many joints of the hand, a wide variety of hand shapes can be characterized as a weighted combination of just two or three main patterns of covariation in joint rotations, or "postural synergies." The present study sought to align muscle synergies with these main postural synergies and to describe the form of membership of motor units in these postural/muscle synergies. Seventeen joint angles and the electromyographic (EMG) activities of several hand muscles (both intrinsic and extrinsic muscles) were recorded while human subjects held the hand statically in 52 specific shapes (i.e., shaping the hand around 26 commonly grasped objects or forming the 26 letter shapes of a manual alphabet). Principal-components analysis revealed several patterns of muscle synergy, some of which represented either coactivation of all hand muscles, or reciprocal patterns of activity (above and below average levels) in the intrinsic index finger and thumb muscles or (to a lesser extent) in the extrinsic four-tendoned extensor and flexor muscles. Single- and multiunit activity was generally a multimodal function of whole hand shape. This implies that motor-unit activation does not align with a single synergy; instead, motor units participate in multiple muscle synergies. Thus it appears that the organization of the global pattern of hand muscle activation is highly distributed. This organization mirrors the highly fractured somatotopy of cortical hand representations and may provide an ideal substrate for motor learning and recovery from injury.

Electromyography↗
PubMed2002

The impact of synaptic depression following brain damage: a connectionist account of "access/refractory" and "degraded-store" semantic impairments.

Neuropsychological studies of patients with acquired semantic impairments have yielded two distinct and contrasting patterns of performance in a spoken-word/picture-matching task (Warrington & Cipolotti, 1996). Patients labeled access/refractory are strongly influenced by presentation rate, semantic relatedness of distractors, and repetition, yet they seem relatively unaffected by lexical frequency. Degraded-store patients, on the other hand, are strongly affected by lexical frequency but are less affected by presentation rate, semantic relatedness, or repetition. Our account of these patterns of performance is based on the distinction between two different types of neurological damage: (1) damage to neuromodulatory systems that function to amplify neural signals while suppressing normal refractory-like effects and (2) damage to connections between groups of neurons that encode semantic information and are sensitive to frequency/familiarity. We present a connectionist model that learns to map spoken-word input to semantic representations and that incorporates a particular form of neural refractoriness referred to as synaptic depression, as well as a simple form of neuromodulation. We show that the model is capable of accounting for the contrasting patterns of semantic impairment under these two different forms of damage and, furthermore, demonstrate how it is capable of handling several documented cases that are exceptions to the basic patterns of impairment. Several predictions and limitations of the present model are discussed.

Brain Damage, Chronic↗
PubMedDate not supplied

[Evoked potential asymmetry in the motor cortex of Macaca fascicularis monkeys during instrumental reflex conditioning].

The monkeys were trained to draw a lever by the left or right hand. Cortical motor evoked potentials (MEP) exhibited the highest amplitude in the cortical motor representation of the "working" hand during the period of learning. The extent of interhemispheral asymmetry of the MEP was the same during operation by the left and right hands. After formation of the conditioned reflex, the interhemispheral asymmetry was significantly less during manipulation by the left hand than by the right one. The decrease of asymmetry was due to relatively high amplitude of the MEP in the left hemisphere during instrumental manipulation of both the right and left hands. The data obtained indicate the formation of functional differences in symmetrical motor areas of the cortex in monkeys during their symmetrical instrumental conditioning by the left and right extremities.

Animals↗
PubMed2004

The role of multisensor data fusion in neuromuscular control of a sagittal arm with a pair of muscles using actor-critic reinforcement learning method.

In this study, we consider the role of multisensor data fusion in neuromuscular control using an actor-critic reinforcement learning method. The model we use is a single link system actuated by a pair of muscles that are excited with alpha and gamma signals. Various physiological sensor information such as proprioception, spindle sensors, and Golgi tendon organs have been integrated to achieve an oscillatory movement with variable amplitude and frequency, while achieving a stable movement with minimum metabolic cost and coactivation. The system is highly nonlinear in all its physical and physiological attributes. Transmission delays are included in the afferent and efferent neural paths to account for a more accurate representation of the reflex loops. This paper proposes a reinforcement learning method with an Actor-Critic architecture instead of middle and low level of central nervous system (CNS). The Actor in this structure is a two layer feedforward neural network and the Critic is a model of the cerebellum. The Critic is trained by the State-Action-Reward-State-Action (SARSA) method. The Critic will train the Actor by supervisory learning based on previous experiences. The reinforcement signal in SARSA is evaluated based on available alternatives concerning the concept of multisensor data fusion. The effectiveness and the biological plausibility of the present model are demonstrated by several simulations. The system showed excellent tracking capability when we integrated the available sensor information. Addition of a penalty for activation of muscles resulted in much lower muscle coactivation while keeping the movement stable.

Algorithms↗
PubMed2006

Spatial relational learning persists following neonatal hippocampal lesions in macaque monkeys.

The hippocampus is important for the acquisition of spatial representations of the environment and consequently in contextual memory. This suggests that the neural substrates underlying spatial cognition might be essential for remembering specific life episodes. Indeed, hippocampal lesions prevent spatial relational learning in adult rodents and monkeys, and result in profound amnesia in adult humans. In contrast, we show here that monkeys with neonatal hippocampal lesions learned new spatial relational information. Our experiments suggest that early hippocampal damage leads to functional brain reorganization that enables spatial information to be acquired through the use of brain regions that normally do not subserve this function.

Adaptation, Physiological↗
PubMed2005

Older adults as adaptive decision makers: evidence from the Iowa Gambling Task.

Older adults process emotional information differently than younger adults and may demonstrate less of a negativity bias on cognitive tasks. The Iowa Gambling Task designed by A. Bechara, H. Damasio, D. Tranel, and A. R. Damasio (1997) has been used to examine the integration of emotion and cognition in a risky-choice decision task and may give insight into differences in the decision-making strategies in younger and older adults. Eighty-eight younger adults (18-34 years) and 67 older adults (65-88 years) completed the Iowa Gambling Task. Using a theoretical decomposition of the task designed by J. R. Busemeyer and J. C. Stout (2002), the authors found that both groups were successful at solving the task but used very different strategies that reflected each group's strength. For younger adults, that strength was learning and memory. For older adults, that strength was an accurate representation of wins and losses (valence).

Adaptation, Psychological↗
PubMed2002

Learning an object from multiple views enhances its recognition in an orthogonal rotational axis in pigeons.

In the natural environment, most objects are seen from several different viewpoints. We explored the nature of recognition after training with multiple views and compared it to recognition after training with only one view. Pigeons were taught with either five views or one view of each of four single-geon objects. Pigeons trained with five views responded more accurately to novel views of an object than did pigeons trained with only one view. This result held even when the novel views came from a rotational axis that was orthogonal to the training axis. These results do not accord with recognition processes involving mental rotation or direct interpolation. Pigeons trained with five views may have formed a view-invariant representation [Psychol. Rev. 94 (1987) 115; Vision Res. 39 (1999) 2885]; alternatively, they may have acquired a more detailed shape space of the objects in which to measure object similarity [Representation and recognition in vision, MIT Press, MA, 1999], or learned to attend to a broader range of features of each object [J. Exp. Anal. Behav. 54 (1990) 69].

Animals↗
PubMed2025

Efficient Detection and Characterization of Targets of Natural Selection Using Transfer Learning.

Natural selection leaves detectable patterns of altered spatial diversity within genomes, and identifying affected regions is crucial for understanding species evolution. Recently, machine learning approaches applied to raw population genomic data have been developed to uncover these adaptive signatures. Convolutional neural networks (CNNs) are particularly effective for this task, as they handle large data arrays while maintaining element correlations. However, shallow CNNs may miss complex patterns due to their limited capacity, while deep CNNs can capture these patterns but require extensive data and computational power. Transfer learning addresses these challenges by utilizing a deep CNN pretrained on a large dataset as a feature extraction tool for downstream classification and evolutionary parameter prediction. This approach reduces extensive training data generation requirements and computational needs while maintaining high performance. In this study, we developed TrIdent, a tool that uses transfer learning to enhance detection of adaptive genomic regions from image representations of multilocus variation. We evaluated TrIdent across various genetic, demographic, and adaptive settings, in addition to unphased data and other confounding factors. TrIdent demonstrated improved detection of adaptive regions compared to recent methods using similar data representations. We further explored model interpretability through class activation maps and adapted TrIdent to infer selection parameters for identified adaptive candidates. Using whole-genome haplotype data from European and African populations, TrIdent effectively recapitulated known sweep candidates and identified novel cancer, and other disease-associated genes as potential sweeps.

Selection, Genetic↗
PubMed1999

Spatial knowledge acquisition from maps and from navigation in real and virtual environments.

In this study, the nature of the spatial representations of an environment acquired from maps, navigation, and virtual environments (VEs) was assessed. Participants first learned the layout of a simple desktop VE and then were tested in that environment. Then, participants learned two floors of a complex building in one of three learning conditions: from a map, from direct experience, or by traversing through a virtual rendition of the building. VE learners showed the poorest learning of the complex environment overall, and the results suggest that VE learners are particularly susceptible to disorientation after rotation. However, all the conditions showed similar levels of performance in learning the layout of landmarks on a single floor. Consistent with previous research, an alignment effect was present for map learners, suggesting that they had formed an orientation-specific representation of the environment. VE learners also showed a preferred orientation, as defined by their initial orientation when learning the environment. Learning the initial simple VE was highly predictive of learning a real environment, suggesting that similar cognitive mechanisms are involved in the two learning situations.

Adolescent↗
PubMedDate not supplied

The end of innocence: historiography and representation in the discursive practice of LD.

In this article, I discuss two interrelated sets of challenges that the discursive practice of learning disability (LD) will need to address, namely, issues associated with the development of a historiography of special education and a more complex understanding of representation issues. I use social theory to address these challenges and raise questions the LD field will need to grapple with as we move toward the consolidation of discursive practices.

Child↗
PubMed1998

Relearning sound localization with new ears.

Because the inner ear is not organized spatially, sound localization relies on the neural processing of implicit acoustic cues. To determine a sound's position, the brain must learn and calibrate these cues, using accurate spatial feedback from other sensorimotor systems. Experimental evidence for such a system has been demonstrated in barn owls, but not in humans. Here, we demonstrate the existence of ongoing spatial calibration in the adult human auditory system. The spectral elevation cues of human subjects were disrupted by modifying their outer ears (pinnae) with molds. Although localization of sound elevation was dramatically degraded immediately after the modification, accurate performance was steadily reacquired. Interestingly, learning the new spectral cues did not interfere with the neural representation of the original cues, as subjects could localize sounds with both normal and modified pinnae.

Adaptation, Physiological↗
PubMed1995

Distinct components of spatial learning revealed by prior training and NMDA receptor blockade.

Synaptic plasticity dependent on N-methyl-D-aspartate (NMDA) receptors is thought to underlie certain types of learning and memory. In support of this, both hippocampal long-term potentiation and spatial learning in a watermaze are impaired by blocking NMDA receptors with a selective antagonist D(-)-2-amino-5-phosphonovaleric acid (AP5) or by a mutation in one of the receptor subunits. Here we report, however, that the AP5-induced learning deficit can be almost completely prevented if rats are pretrained in a different watermaze before administration of the drug. This is not because of stimulus generalization, and occurs despite learning of the second task remaining hippocampus dependent. An AP5-induced learning deficit is, however, still seen if the animals are pretrained using a non-spatial task. Thus, despite its procedural simplicity, the watermaze may involve multiple cognitive processes with distinct pharmacological properties; although required for some component of spatial learning, NMDA receptors may not be required for encoding the spatial representation of a specific environment.

2-Amino-5-phosphonovalerate↗
PubMed2004

Phonemic manipulation in Japanese: an fMRI study.

Phonological awareness is the ability to manipulate abstract phonological representations of language and is crucial to the process of learning to read. The neural substrates underlying this appear to be modality-independent at least in alphabetic languages. Japanese language has different orthographic "kana" system, in which each "kana" character strictly corresponds to a syllable. To investigate the neural substrates underlying phonological manipulation of the Japanese language, functional magnetic resonance imaging (fMRI) was used. Neuroimaging data were obtained from adult healthy volunteers during auditory and visual vowel exchange tasks, identical except for the modality of stimuli presentation: a voice and Japanese "kana" characters. Cerebellar vermis was activated by vowel exchange tasks of both modalities. The posterior parts of the superior temporal sulcus (STS) were active during the auditory tasks, suggesting that phonological representations of auditory stimuli are manipulated in this area. These findings are consistent with the previous studies with alphabetic languages. In contrast, the intraparietal sulci, which has been implicated for visuospatial tasks, was active during the visual tasks. This modality-dependent activation may indicate that the simple orthographic rule of the Japanese allows an alternate visual strategy to conduct the phonological awareness task, bypassing manipulation of phonological representation.

Adult↗
PubMed2002

Coordination dynamics of learning and transfer across different effector systems.

If different effector systems share a common task-specific coordination dynamics, transfer and generalization of sensorimotor learning are predicted. Subjects learned a visually specified phase relationship with either the arms or the legs. Coordination tendencies in both effector systems were evaluated before and after practice to detect attractive states of the coordination dynamics. Results indicated that learning a novel relative phase with a single effector system spontaneously transferred to the other, untrained effector system. Transfer was revealed not only as improvements in performance but also as modifications of each system's initial (prelearning) coordinative landscape. What is learned, appears to be a high-level but neurally instantiated dynamic representation of skilled behavior that proves to be largely effector independent, at least across anatomically symmetric limbs.

Adult↗
PubMed2006

Dynamics of memory representations in networks with novelty-facilitated synaptic plasticity.

The ability to associate some stimuli while differentiating between others is an essential characteristic of biological memory. Theoretical models identify memories as attractors of neural network activity, with learning based on Hebb-like synaptic modifications. Our analysis shows that when network inputs are correlated, this mechanism results in overassociations, even up to several memories "merging" into one. To counteract this tendency, we introduce a learning mechanism that involves novelty-facilitated modifications, accentuating synaptic changes proportionally to the difference between network input and stored memories. This mechanism introduces a dependency of synaptic modifications on previously acquired memories, enabling a wide spectrum of memory associations, ranging from absolute discrimination to complete merging. The model predicts that memory representations should be sensitive to learning order, consistent with recent psychophysical studies of face recognition and electrophysiological experiments on hippocampal place cells. The proposed mechanism is compatible with a recent biological model of novelty-facilitated learning in hippocampal circuitry.

Animals↗
PubMed1991

Hemisphere differences in conditional learning: an ERP-study.

The present ERP-experiment investigated brain asymmetry in encoding and representation of the formation of conditional associations in a Pavlovian framework. Two consonant-vowel syllables were used as conditional stimuli, one of them paired with an unconditional noise during the acquisition phase (CS+), the other never paired with noise (CS-). During a subsequent test phase, both CS cues were presented dichotically i.e. at the same time. Half of the subjects had the CS+ probe presented to the right ear (contralateral to the left hemisphere). The other half of the subjects had the CS+ presented to the left ear (contralateral to the right hemisphere). ERPs from F3, Fz and F4 leads were recorded. The paradigm used was an adaptation of the standard dichotic listening technique for the study of learned associations. The results showed that conditional associative learning occurred during the acquisition phase in both "groups". The P235 latency was furthermore delayed over the left hemisphere to the CS+ probe. During the dichotic test phase, there was a clear laterality or asymmetry effect between the groups for both the N125 and N450 components, possibly supporting an interpretation of asymmetry in encoding and cortical representation of a conditional association, although alternative interpretations remain open.

Adult↗
PubMed2000

Learning to write letters: transfer in automated movements indicates modularity of motor programs in human subjects.

Many automatic movements are open-loop, feed-forward motor programs (MP) that are kinematically well characterized by smooth speed and acceleration curves. However, it is unclear whether their internal representation consists of monolithic blocks or subroutines. This question was investigated using a learning paradigm of a writing task. Fifty-nine normal subjects were presented with two similar, but different new letters. Every subject practiced each letter in a series of 60 trials, with the order of letter series randomized. Every session was continuously recorded by a digitizing tablet. Using kinematic analysis, we measured the number of vertical acceleration peaks as an indication of the number of corrective movements (COM). Since COM declined as automatization was approached, we could quantitatively infer progress in motor learning under natural learning conditions. In the case of modular storage of MP, transfer in-between letters was expected due to the re-use of pre-learned motor subroutines. Statistical analysis showed that the exponential model described the data much better than the linear model (residual error: P<0.88 and P<0.00001, respectively), as expected for a learning paradigm. There was no difference between letters per se (P<0.77). Motor improvement differed significantly (P<0.02) between the first and the second series; there was a much greater reduction of COM in the second series (50.1 vs. 41.1%). This difference can be logically ascribed to transfer, indicating that automated movements are stored in motor subroutines.

Biomechanical Phenomena↗
PubMed2001

Predicting spontaneous recovery of memory.

Long after a new language has been learned and forgotten, relearning a few words seems to trigger the recall of other words. Neural-network models indicate that this form of spontaneous recovery may result from the storage of distributed representations, which are thought to mediate human memory. Here we use a psychomotor learning task to show that a corresponding effect of spontaneous memory recovery occurs in human subjects.

Humans↗