Large-scale brain models

A main focus of my current research is the modeling of integrative brain systems. Integrative brain systems are multi-component biological-scale neural models that include analogs of perceptual, emotional, cognitive, planning, or motor brain areas to control adaptive behavior in virtual or robotic agents. Building this system is impossible without solid research capturing mechanisms at multiple scales, from plasticity in synapses to long-range communication across brain areas. However, in order to mechanistically understand learning and behavior in adaptive agents, and transfer these mechanisms to technological applications, all these components should be integrated in an embodied neural system capable of perception-cognition-emotion-action cycles while interacting with a real or virtual environment.

An apparently simple behavior, for example a fetching task where a human asks a robotic assistant to pick up a water bottle in the kitchen, illustrates this complexity. Mechanisms ranging from auditory localization, speaker identification, early auditory filtering, to speech perception and language understanding involving multiple cortical and subcortical brain regions need to be coordinated with motivational and planning systems to appropriately arrange a sequence of motor actions that lead to the water bottle. These plan needs to interact with cortical and subcortical representations responsible for spatial navigation, while at the same time the visual system coordinates complementary properties in the what and where system to localize and recognize objects and landmarks, and guide reaching movements to the water bottle. While the plan is generated and carried out, unexpected events such as moving object, a child playing in the path of the robot, or a changed spatial layout in the house need to trigger reactive avoiding behaviors and update the plan, while keeping the overall goal in short term memory.

My research focuses on a system approach where all these components are co-designed and integrated to provide a way to tackle science and technology problems. As director of the Boston University Neuromorphics Lab, and in collaborations with several academic and industrial partners, I have pioneered a system approach in building a whole-brain system is called MoNETA (Modular Neural Exploring Traveling Agent).

MoNETA consists of different brain modules (32M neurons, 13B synapses), ranging from sensory, to motivation, to planning and navigation areas interfaced either with a virtual environment or with a robotic platform. MoNETA has been tested in a virtual environment where it learns to perform a virtual environment version of the Morris Water Maze task, and has been embodies in robotic platforms where it performs spatial navigation tasks.

MoNETA and its evolution will be the basis of current and future technological applications in robotics and business intelligence, in collaboration with iRobot and Hewlett-Packard (HP).

The first version of MoNETA has been completed in the DARPA SyNAPSE project, where MoNETA was able to successfully negotiate the Morris Water Maze task.
♦ For additional information, visit the Neuromorphics Lab website for completed projects and current projects MoNETA pages.

 

Articles

  • Ames, H. Mingolla, E., Sohail, A., Chandler, B., Gorchetchnikov, A., Léveillé, J., Livitz, G. and Versace, M. (2012) The Animat. IEEE Pulse, January/February 2012. PDF
  • Livitz G., Versace M., Gorchetchnikov A., Vasilkoski Z., Ames H., Chandler B., Leveille J. and Mingolla E. (2011) Scalable adaptive brain-like systems, The Neuromorphic Engineer: 10.2417/1201101.003500 February 2011. PDF
  • Versace M. and Chandler B. (2010) MoNETA: A Mind Made from Memristors. IEEE Spectrum, December 2011. PDF

Abstracts & Conference papers

  • Gorchetchnikov A, Versace M, Ames H, Chandler B, Leveille J, Livitz G, Mingolla E, Snider G, Amerson R, Carter D, Abdalla H, and Qureshi MS. (2011). Review and unification of learning framework in Cog Ex Machina platform for memristive neuromorphic hardware. In: Proceedings of the International Joint Conference on Neural Networks, number 531 in IEEE CD-ROM Catalog Number: CFP11IJS-CDR, ISBN: 978-1-4244-9636-5 pp. 2601–2608. PDF
  • Leveille J., Ames H., Chandler B., Gorchetchnikov A., Mingolla E., Patrick S. and Versace M. (2010) Learning in a distributed software architecture. To appear in Lecture Notes for Computer Sciences, Social Informatics, and Telecommunications Engineering (LNICST). PDF
  • Livitz G, Ames H, Chandler B, Gorchetchnikov A, Leveille J, Vasilkoski Z, Versace M, Mingolla E, Snider G, Amerson R, Carter D, Abdalla H, and Qureshi MS. (2011). Visually-Guided Adaptive Robot (ViGuAR). In: Proceedings of the International Joint Conference on Neural Networks, number 620 in IEEE CD-ROM Catalog Number: CFP11IJS-CDR, ISBN: 978-1-4244-9636-5 pp. 2944–2951. PDF
  • Vasilkoski Z, Ames H, Chandler B, Gorchetchnikov A, L´veill´ J, Livitz G, Mingolla E, and Versace M. (2011). Review of stability properties of neural plasticity rules for implementation on memristive neuromorphic hardware. In: Proceedings of the International Joint Conference on Neural Networks, number 524 in IEEE CD-ROM Catalog Number: CFP11IJS-CDR, ISBN: 978-1-4244-9636-5 pp. 2563–2569. PDF
  • Gorchetchnikov A., Leveille J., Versace M., Ames H., Livitz G, Chandler B., Mingolla E., Carter D., Amerson R., and Snider G. (2011). MoNETA: Massive parallel application of biological models navigating through virtual Morris water maze and beyond. Computational Neuroscience Meeting Abstracts, Stockholm, Sweden (CNS 2011)
  • Leveille, J., Livitz, G., Ames, H., Chandler, B., Gorchetchnikov, A., Versace, M., Mingolla, E. and Snider, G. (2011). Learning to see in a virtual world. Neuroinformatics 2011, Boston, MA.
  • Leveille J., Ames H., Chandler B., Gorchetchnikov A., Livitz G., Versace M. and Mingolla E. (2011) Invariant object recognition and localization in a virtual animat. International Conference on Cognitive and Neural Systems (ICCNS) 2011, Boston, MA, USA.
  • Leveille J., Ames H., Chandler B., Gorchetchnikov A., Livitz G., Versace M. and Mingolla E. (2011) Object recognition and localization in a virtual animat: large-scale implementation in dense memristive memory devices. International Joint Conference on Neural Networks (IJCNN) 2011, San Jose, CA, USA.
  • Gorchetchnikov A., Ames H., Chandler B., Leveille J., Livitz G., Mingolla E. and Versace M. (2010) MoNETA: Modular Neural Exploring Traveling Agent. Functional Connections workshop, CELEST, Boston, October 2010.

Collaborators

I am the leader of the Neuromorphics Lab, a highly collaborative lab with connections across both academia and industry.