Computational challenges & opportunities P. Perona California - - PowerPoint PPT Presentation

▶
computational challenges opportunities
SMART_READER_LITE
LIVE PREVIEW

Computational challenges & opportunities P. Perona California - - PowerPoint PPT Presentation

Aug 29, 2022 202 likes •479 views

Computational challenges & opportunities P. Perona California Institute of Technology 4 December 2014 Press alarm? Time is money ER worse error rate better response time RT $ feeding frienzy picking apples t Tuned neurons

slide-1
SLIDE 1

Computational challenges &

  • pportunities
  • P. Perona

California Institute of Technology 4 December 2014

slide-2
SLIDE 2
slide-3
SLIDE 3

Press alarm?

slide-4
SLIDE 4

RT ER

response time error rate worse better

Time is money

slide-5
SLIDE 5
slide-6
SLIDE 6
slide-7
SLIDE 7

t

$

picking apples feeding frienzy

slide-8
SLIDE 8

Tuned neurons

[Hubel & Wiesel]

  • rientation

and frequency

[Graf and Movshon 2012]

slide-9
SLIDE 9

Principles of model selection

  • Optimal observer
  • Occam’s razor
slide-10
SLIDE 10

Wald’s sequential analysis

t

(0, T) = I0 [ t1 [ I1 [ t2 [ · · ·

log R(Ik) = log P(n = 0|C = 2, t ∈ Ik) P(n = 0|C = 1, t ∈ Ik) = log ΠN

i=1P(n = 0|λi 2∆tk)

ΠN

i=1P(n = 0|λi 1∆tk) = ∆tk N

X

i=1

(λi

1 − λi 2)

log R(tk) = lim

δt→0 log P(n = 1|λi(k) 2

δt) P(n = 1|λi(k)

1

δtk) = lim

δt→0 log (λi(k) 2

δt)1e−λi(k)

2

δt

(λi(k)

1

δt)1e−λi(k)

1

δt = log λi(k) 2

λi(k)

1

t1 t2 tk I1 I2 Drift Action potential jumps

slide-11
SLIDE 11

Model

z’

S S S L S S L S

Bow

... ...

hypercolumn:

λmax λmin = 1 N = 32

∆θ = 250

T1 T0

decision

slide-12
SLIDE 12

Predictions vs data

slide-13
SLIDE 13

Subjects’ parameters

slide-14
SLIDE 14

Technology

slide-15
SLIDE 15

Summary (1)

  • Perception in the context of behavior
  • Delay vs error
  • Account for action potentials
  • Ideas for technology?
slide-16
SLIDE 16

What is behavior?

Marla B. Sokolowski

Nature Reviews Genetics 2, 879-892 (2001)

slide-17
SLIDE 17

From images to actions

x y t Video stream tracking Trajectory

I(x, y, t)

⇤ x(t)

classification eat drink mate Ethogram dynamics t1 t2 10Mb/s 10kb/s 1b/s

(t1, b1) . . . (tN, bN)

slide-18
SLIDE 18

[Branson et al. Nature Methods, Jun. ’09]

slide-19
SLIDE 19
slide-20
SLIDE 20
slide-21
SLIDE 21

Phenotyping

[Dankert et al., Nature Methods, 2009]

slide-22
SLIDE 22

Ethograms

[Dankert et al., Nature Methods, April 2009]

slide-23
SLIDE 23

Center 1 Center 2 Center 3 Center 4 Center 5 Center 6 Center 7 Center 8 Center 9 Center 10 Center 11 Center 12 Center 13 Center 14 Center 15 Center 16 Center 17 Center 18 Center 19 Center 20 Center 21 Center 22 Center 23 Center 24 Center 25 Center 26 Center 27 Center 28 Center 29 Center 30 Center 31 Center 32 Center 33 Center 34 Center 35 Center 36 Center 37 Center 38 Center 39 Center 40 Center 41 Center 42 Center 43 Center 44 Center 45 Center 46 Center 47 Center 48 Center 49 Center 50 Center 51 Center 52 Center 53 Center 54 Center 55 Center 56 Center 57 Center 58 Center 59 Center 60 Center 61 Center 62 Center 63 Center 64

Unsupervised analysis

23

slide-24
SLIDE 24

1 5 2 6 4 3 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 Dendrogram of distances 1 2 3 4 5 6 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 mean distance from i i dist 1 2 3 4 5 6 0.8 1 1.2 1.4 1.6 1.8 dist i Percentile distance from i

Behaviors that stand out

slide-25
SLIDE 25

Summary (2)

  • Genes -> Brains -> Behavior
  • Blue-collar engineering work
  • Unsupervised discovery
slide-26
SLIDE 26

Grand challenges

  • Genes -> Brains -> Behavior
  • Theory of behavior (info, control)
  • Spike-based computation