Generalization in vision and motor control

By Tomaso A. Poggio, Emilio Bizzi. Nature, 2004

378 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/aa482af08f37c1b5604d2821246878b10e578c7c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

Citations in the corpus

Trade-Off between Object Selectivity and Tolerance in Monkey Inferotemporal Cortex (2007, methodology): “Because AND-like, approximately Gaussian tuning functions can be implemented through divisive normalization (Poggio and Bizzi, 2004), these simulations show the consistency between the observed trade-off and divisive normalization models (Heeger, 1992; Carandini et al., 1997; Recanzone et al.,…”
A quantitative theory of immediate visual recognition. (2007, background): “Stages of simple cells with Gaussian-like tuning (bold circles and arrows), which provide generalization (Poggio and Bizzi, 2004), are interleaved with layers of complex units (dashed circles and arrows), which perform a MAX-like operation on their inputs and provide invariance to position and…”
Motor Task Variation Induces Structural Learning (2009, neutral): “Typically, learning the mapping from sensory inputs X to a motor command U has been conceptualized as a mapping U = P i wigiðXÞ, where gið,Þ are so-called basis functions and wi are the adjustable parameters of the mapping [2, 6, 15, 16].”
Unsupervised Learning of Visual Features through Spike Timing Dependent Plasticity (2007, methodology): “Supervised learning at this stage involves adjusting the synaptic weights c so as to minimize a (regularized) error on the training set [27].”
Robust Object Recognition with Cortex-Like Mechanisms (2007, background): “This is consistent with well-known response properties of neurons in primate inferotemporal cortex and seems to be the key property for learning to generalize in the visual and motor systems [42].”
How the brain might work: a hierarchical and temporal model for learning and recognition (2008, background): “In [79], Poggio clearly identifies a connection between generalization and hierarchy and argues that generalization will depend more on the architecture than on the detailed nature of the algorithms.”
A theory for how sensorimotor skills are learned and retained in noisy and nonstationary neural circuits (2013, background): “However, surely the motor system need not—and from a computational efficiency perspective should not (41)—explicitly represent all aspects of the motor behaviors it can perform.”
Biological pattern generation: the cellular and computational logic of networks in motion. (2006, background): “The output of the activity captures essential features of the locomotor output. d´Avella and Bizzi, 2005; Poggio and Bizzi, 2004).”
The neural origin of muscle synergies (2013, background): “In broader terms, muscle synergies may be essential components in the architecture of the motor system that allow generalization to occur (Poggio and Bizzi, 2004).”
Generalization and similarity in exemplar models of categorization: Insights from machine learning (2008, methodology): “These nets have repeatedly been advocated by Poggio and coworkers (Poggio, 1990; Poggio & Bizzi, 2004 ) as a plausible model for brain function.”
Toward an Integration of Deep Learning and Neuroscience (2016, methodology): “This relies on a theory of generalization in learning based on combinations of tuned units (Poggio and Bizzi, 2004), which has been applied to both vision and motor control.”
A Canonical Neural Circuit for Cortical Nonlinear Operations (2008, background): “In the motor system, the activity of a spinal cord neuron can be regarded as being tuned to a particular pattern of force fields or limb movements (Poggio & Bizzi, 2004).”
A Theory of Object Recognition: Computations and Circuits in the Feedforward Path of the Ventral Stream in Primate Visual Cortex (2005, methodology): “We also replaced the final classification stage (equivalent to an RBF scheme [Poggio and Bizzi, 2004] in the model, which includes the S4 units in IT and the task-specific circuits from IT to PFC) with a more standard “computer vision classifier” (Ada boost).”
Visual image reconstruction from human brain activity using a combination of multiscale local image decoders. (2008, background): “A large variety of motor actions could be described by a combination of putative modules (Poggio and Bizzi, 2004).”
Task-Specific Generalization of Discrete and Periodic Dynamic Movement Primitives (2010, background): “example movement trajectories are stored directly in memory [30], [31].”