Hierarchical models of object recognition in cortex

By Maximilian Riesenhuber, Tomaso A. Poggio. 1999

Subfields: Computer Vision and Pattern Recognition, Neuroscience

How researchers see this paper

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

In Computer Vision and Pattern Recognition

In Neuroscience

Citations in the corpus

BAM: Bottleneck Attention Module (2018, neutral): “More visualizations and analysis are included in the supplementary material due to space constraints.”
Deep learning in neural networks: An overview (2014, methodology): “The neurophysiologically plausible topology of the feedforward HMAX model (Riesenhuber & Poggio, 1999) is very similar to the one of the 1992 Cresceptron (and thus to the 1979 Neocognitron).”
Reluplex: An Efficient SMT Solver for Verifying Deep Neural Networks (2017, background): “Artificial neural networks [7, 31] have emerged as a promising approach for creating scalable and robust systems.”
View-tolerant face recognition and Hebbian learning imply mirror-symmetric neural tuning to head orientation (2016, positive): “Several hierarchical models of object recognition [17, 40, 43, 9] and face recognition [8, 29, 14] feature a progression from view-specific early processing stages to view-invariant later processing stages similar to ML/MF and AM, respectively.”
Lifelong learning of human actions with deep neural network self-organization (2017, positive): “Mechanisms of MAX-pooling in hierarchical models are also compatible with neurophysiological data (Riesenhuber & Poggio, 1999), e.g. showing that the response of neurons in the inferotemporal (IT) cortex is dominated by the stimulus producing the highest firing rate (Sato, 1989).”
The global landscape of cognition: hierarchical aggregation as an organizational principle of human cortical networks and functions (2015, methodology): “General algorithmic procedure for plotting and sorting the behavioral databases upon connectivity matrices can be outlined in four steps: (1) Compute connectivity order, (2) Compute behavioral element frequencies, (3) Bin connectivity ordering, (4) Regress behavioral element frequency across bins, (”
High-Performance Neural Networks for Visual Object Classification (2011, positive): “…in conjunction with experimental studies of the visual cortex justify the use of such ﬁlters in the so-called standard model for object recognition (Riesenhuber and Poggio, 1999; Serre et al. , 2007; Mutch and Lowe, 2008), whose ﬁlters are ﬁxed, in contrast to those of Convolutional Neural…”
How does the brain solve visual object recognition? (2012, positive): “This can produce a gradual increase of the tolerance to variation in object appearance along the hierarchy (e.g., Fukushima, 1980; Riesenhuber and Poggio, 1999b; Serre et al., 2007a).”
Group Invariant Scattering (2011, neutral): “Applying it to g = U [p]f − U [p]h together with U [p]f ⋆ φ2J = SJ [p]f and U [p]f ⋆ ψ2J r = U [p + 2 Jr]f gives ‖SJ [p]f − SJ [p]h‖2 = ‖SJ+1[p]f − SJ+1[p]h‖2 (42) + ∑ r∈G+ ‖U [p + 2Jr]f − U [p + 2Jr]h‖2 Since SJ+1[PJ+1]U [p+2Jr]f = {SJ+1[p+2Jr+p′′]}p′′∈PJ+1 , and SJ+1[PJ+1]f is nonexpansive, it…”
Sum-product networks: A new deep architecture (2011, background): “(Cf. Riesenhuber and Poggio [20], which also uses max and sum nodes, but is not a probabilistic model.)”
Learned-Norm Pooling for Deep Feedforward and Recurrent Neural Networks (2013, background): “A pooling operator, in this context, is understood as a way to summarize a high-dimensional collection of neural responses and produce features that are invariant to some variations in the input (across the ﬁlter outputs that are being pooled).”
On Invariance and Selectivity in Representation Learning (2015, methodology): “Hierarchical representations, based on multiple layers of computations, have naturally arisen from models of information processing in the brain [15,33].”
Bridging the Gaps Between Residual Learning, Recurrent Neural Networks and Visual Cortex (2016, background): “Most of the models of cortex that led to the Deep Convolutional architectures and followed them – such as the Neocognitron [6], HMAX [26] and more recent models [1] – have neglected the layering in each cortical area and the feedforward and recurrent connections within each area and between them.”
Distributed optimization of deeply nested systems (2012, background): “1), which were originally inspired by biological systems such as the visual and auditory cortex in the mammalian brain (Riesenhuber and Poggio, 1999; Serre et al., 2007; Gold and Morgan, 2000), and which have been proven very successful at learning sophisticated tasks, such as recognizing faces or…”
A High-Throughput Screening Approach to Discovering Good Forms of Biologically Inspired Visual Representation (2009, positive): “[1–6]), biologically-inspired models are routinely among the highest-performing artificial vision systems in practical tests of object and face recognition [7–12].”
A Biologically Inspired System for Action Recognition (2007, methodology): “It extends an earlier neurobiological model of motion processing in the dorsal stream of the visual cortex by Giese & Poggio [10].”
The Distributed Nature of Working Memory. (2017, background): “The posterior-tofrontal gradient of increasing abstraction proposed here (see ‘Different Levels of Abstraction’) can be interpreted as a form of complexity reduction [189] or ‘chunking’ [153].”