BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:From the olfactory cocktail party to markerless tracking -  Alexan
 der Mathis\, Department of Molecular and Cellular Biology\, Harvard Univer
 sity
DTSTART:20171026T090000Z
DTEND:20171026T100000Z
UID:TALK94201@talks.cam.ac.uk
CONTACT:Prof Máté Lengyel
DESCRIPTION:The olfactory system\, like other sensory systems\, can detect
  specific stimuli of interest amidst complex\, varying backgrounds. To gai
 n insight into the neural mechanisms underlying this ability\, we estimate
 d a model for mixture responses that incorporated nonlinear interactions a
 nd trial-to-trial variability and explored potential decoding mechanisms t
 hat can mimic mouse performance when given glomerular responses as input. 
 We find that a linear decoder could match mouse performance using just a s
 mall subset of the glomeruli. However\, when such a decoder is trained onl
 y with single odors\, it generalizes poorly to mixture stimuli. We show th
 at mice similarly fail to generalize\, suggesting that they learn this seg
 regation task discriminatively by adjusting task-specific decision boundar
 ies without taking advantage of a demixed representation of odors (Mathis 
 et al. 2016). I will present ongoing experiments designed to challenge thi
 s model\, for which mice were trained in a semisupervised fashion.\n\nMoti
 vated by the weak constraints for elucidating the neural mechanisms in thi
 s olfactory perception task\, we are increasingly turning our attention to
  more challenging behaviors – like trail tracking. Mice naturally follow
  odor trails and one can easily gather large amounts of video data. Howeve
 r\, reliably extracting particular aspects of a behavior\, like the positi
 on of the snout\, can be difficult. In motor control studies reflective ma
 rkers are often used to assist with computer-based tracking\, yet markers 
 are highly intrusive for smaller animals. I will present a Deep Learning b
 ased method for markerless tracking and demonstrate the versatility of thi
 s framework in three different tasks: trail-tracking\, social behaviors an
 d skilled forelimb reaching. This algorithm is trained in an end-to-end fa
 shion based on training data with labels for specific anatomical points of
  interest. Crucially\, only a small set of frames is required for training
  and the algorithm generalizes to test data in a quantitatively comparable
  way to human annotators. \n
LOCATION:Cambridge University Engineering Department\, CBL\, BE4-38 (http:
 //learning.eng.cam.ac.uk/Public/Directions)
END:VEVENT
END:VCALENDAR