Counting objects with biologically inspired regulatory-feedback networks

Neural networks are relatively successful in recognizing individual patterns. However, when images consist of combination of patterns, a preprocessing step of segmentation is required to avoid combinatorial explosion of the training phase. In practical applications, segmentation is a context dependent task which itself requires recognition. In this paper we propose and develop a biologically inspired neural architecture that can recognize and count an arbitrary collection of objects even if trained with individual objects, without making use of additional segmentation algorithms. The two essential features that govern the neurons in this algorithm are 1. dynamical feedback and 2. competition for activation. We show analytically that while the equations governing the output neurons are highly nonlinear in individual feature amplitudes, they are linear in groups of feature amplitudes. We further demonstrate through simulations, that our architecture can precisely count and recognize scenes in which three and four non-overlapping patterns are presented simultaneously. The ability to generalize numerosity outside the training distribution with a simple learning scheme, lack of connection weights and segmentation algorithms prove regulatory feedback networks not only beneficial for machine learning tasks but also for biological modeling of animal vision.

[1]  D. Kunz Possibilities and Limitations of Automated Platelet Counting Procedures in the Thrombocytopenic Range , 2001, Seminars in thrombosis and hemostasis.

[2]  Dmitri A. Rachkovskij,et al.  Binding and Normalization of Binary Sparse Distributed Representations by Context-Dependent Thinning , 2001, Neural Computation.

[3]  Eyal Amir,et al.  Input Feedback Networks: Classification and Inference Based on Network Structure , 2008, AGI.

[4]  Stephen Grossberg,et al.  A neural model of how the brain represents and compares multi-digit numbers: spatial and categorical processes , 2003, Neural Networks.

[5]  Eyal Amir,et al.  Shedding weights: More with less , 2008, 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence).

[6]  A. Kratz,et al.  Effects of yeast on automated cell counting. , 2006, American journal of clinical pathology.

[7]  Dražen Domijan A neural model of quantity discrimination , 2004, Neuroreport.

[8]  A Treisman,et al.  Feature analysis in early vision: evidence from search asymmetries. , 1988, Psychological review.

[9]  James A. Reggia Virtual Lateral Inhibition in Parallel Activation Models of Associative Memory , 1985, IJCAI.

[10]  Khurshid Ahmad,et al.  A competitive neural model of small number detection , 2006, Neural Networks.

[11]  B Swolin,et al.  Differential counting of blood leukocytes using automated microscopy and a decision support system based on artificial neural networks--evaluation of DiffMaster Octavia. , 2003, Clinical and laboratory haematology.

[12]  C. V. D. Malsburg,et al.  Frank Rosenblatt: Principles of Neurodynamics: Perceptrons and the Theory of Brain Mechanisms , 1986 .

[13]  Frank Rosenblatt,et al.  PRINCIPLES OF NEURODYNAMICS. PERCEPTRONS AND THE THEORY OF BRAIN MECHANISMS , 1963 .

[14]  Tsvi Achler,et al.  Input shunt networks , 2002, Neurocomputing.

[15]  From Humble Neural Beginnings Comes Knowledge of Numbers , 2003, Neuron.

[16]  Tim Holmes,et al.  Object segmentation, quantification, counting, and tracking , 2004, SPIE BiOS.

[17]  James A. Reggia,et al.  A Competitive Distribution Theory of Neocortical Dynamics , 1992, Neural Computation.

[18]  E. Spelke,et al.  Language and Conceptual Development series Core systems of number , 2004 .

[19]  S. Dehaene,et al.  Interactions between number and space in parietal cortex , 2005, Nature Reviews Neuroscience.

[20]  Frank E. McFadden,et al.  Convergence of competitive activation models based on virtual lateral inhibition , 1995, Neural Networks.

[21]  Tsvi Achler,et al.  Object classifiction with recurrent feedback neural networks , 2007, SPIE Defense + Commercial Sensing.