SlideShare a Scribd company logo
Learning Anticipation via Spiking Networks:
Application to Navigation Control
Paolo Arena et al.
2009
IEEE Transactions on
Neural Network
Navigation Techniques in Robotics
Spatial Information Processing
▪ Rat has the navigation cells that represent the place[1], grid[2],
boundary[3] and direction[4] information in the environment
[1] O’keefe and Dostrovsky, 1971, Brain Res [3] Lever et al., 2009, J Neurosci. [5] Moser et al., 2008, Annu Rev Neurosci
[2] Fyhn et al., 2004, Science [4] Taube et al., 1990, J Neurosci. [6] Zhang et al., 2013, Philos Trans R Soc Lond B Biol Sci.
Place cell
[5]
Head direction ( °)
Firingrate(Hz)
Preferred direction
Head direction cell
[4]
Boundary vector cell
[7]
Grid cell
[5]
Rat trajectory
Cell spikes
Neural Network Architecture
▪ Artificial neural network
▫ Output is calculated by
summation of inputs with
weight vectors
▫ The weight vector is
modulated to get the right
output
▪ Spiking neural network
▫ Output is the result of
neural activity by
presynaptic inputs
▫ The synaptic weight is
modulated by biological
learning rule
Spike-timing Dependent Plasticity (STDP)
▪ Excitatory postsynaptic current (EPSC) depends on the relative
timing of pre- and postsynaptic action potentials
▪ Unsupervised learning rule
[7] Bi and Poo, 1998, J Neurosci.
Δt (tpost – tpre)
ChangeinEPSCamplitude(%)
Pre
Post
[7]
Learning Factors
▪ Classical conditioning provided the source of inspiration for
the implementation of navigation task algorithm
Unconditioned stimuli Unconditioned response
Conditioned stimuli Conditioned response
Method
▪ The Spiking Network Model
▫ Sensory neurons are neurons connected to sensors
▫ Motor neurons drive robot motors
▫ Interneurons
▫ Each single neurons are modeled by Izhikevich model
v: membrane potential
u: recovery variable
I: input current
[8] Izhikevich, IEEE, 2004
20 ms
a = 0.02
b = -0.1
c = -55
d = 6
i j
[8]
Method
▪ STDP modification function
Δt (tpre – tpost)
ഥ𝒈 𝒎𝒂𝒙
𝜏+ = 20
𝜏− = 10
𝐴+ = 0.02
𝐴− = −0.02
[9] Song et al., 2000, Nat Neurosci.
[9]
Method
▪ Robot simulation
▫ Dual-drive wheeled virtual robot (right motor: RM, left motor: LM)
▫ Two collision sensor (US)
▫ Two rand finder sensor ([0°, -45°], [0°, 45°], CS)
▫ Two proximity sensor (for target)
Left range
finder sensor
Right range
finder sensor
Range of
proximity sensor Camera view (black line)
Method
▪ Robot experiment
▫ Real roving robot
▫ Lynxmotion 4WD2 Robot
▫ Four infrared distance sensors (14 to 80 cm)
▫ Collision (threshold at 23 cm)
▫ RF04 USB radio telemetry module
Method
▪ Network for obstacle avoidance
▫ The robot moves (0.3 r.u. for each simulation step, time window: 300 ms)
in an environment filled with randomly placed obstacles
Collision sensor
Method
Proximity target sensor
▪ The network for target approaching
▫ The barycenter of red object is provided to the network of nine neurons
as a visual input
Method
▪ The network for navigation with visual cue
Result – Obstacle Avoidance
▪ Obstacle avoidance
Robot is driven by UR
Robot is driven by CR
Robot is driven by CR
Obstacleavoidancenumber
Result - Obstacle Avoidance
▪ The number of collision and
wrong decision is decreased by
STDP learning rule
NumberofcollisionDistancebetweenrobotand
closestobstacles
Result – Synaptic Weight Change
▪ Obstacle avoidance
Potentiation
Potentiation
Depression
Depression
Result – Obstacle Avoidance
▪ Obstacle avoidance
▫ Trajectories obtained in the case in which the robot faces one obstacle
Before learning End of learning phase
Result – Synaptic Weight Change
▪ Obstacle avoidance
▫ Robot experiments
Result – Target Approaching
▪ Target approaching
Without learning
After learning
(without visual input)
(with visual input)
Long run simulation
Range of
proximity sensor
Result – Target Approaching
(without visual input)
(with visual input)
Before learning
End of learning phase
Two targets
Result - Navigation with Visual Cue
Cases of training environments
Markers
Case (a)
Solving maze
based on
case (a)
Conclusion
▪ Biological spiking neural network shows the efficiency for the
dynamic systems instead of a static relationship between input
and output
▪ STDP unsupervised learning rule is one of a simple way for
navigation algorithm
▪ The spiking cells are suitable for investigate the temporal code
of neural activity
Discussion
▪ The design of neural network is not considered for
neuromorphic
▪ These networks cannot recall the specific spatial memory
(place)

More Related Content

What's hot

ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSIONADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
ijistjournal
 
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
cscpconf
 
Microscopy images segmentation algorithm based on shearlet neural network
Microscopy images segmentation algorithm based on shearlet neural networkMicroscopy images segmentation algorithm based on shearlet neural network
Microscopy images segmentation algorithm based on shearlet neural network
journalBEEI
 
Efficient Neural Network Architecture for Image Classfication
Efficient Neural Network Architecture for Image ClassficationEfficient Neural Network Architecture for Image Classfication
Efficient Neural Network Architecture for Image Classfication
Yogendra Tamang
 
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
sipij
 
automated perimeters
automated perimetersautomated perimeters
automated perimeters
gueste49b03
 
Image recognition
Image recognitionImage recognition
Image recognition
Stig-Arne Kristoffersen
 
Ag044216224
Ag044216224Ag044216224
Ag044216224
IJERA Editor
 
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
ijcsit
 
Random Neural Network (Erol) by Engr. Edgar Carrillo II
Random Neural Network (Erol) by Engr. Edgar Carrillo IIRandom Neural Network (Erol) by Engr. Edgar Carrillo II
Random Neural Network (Erol) by Engr. Edgar Carrillo II
Edgar Carrillo
 
The 5th WBA Hackathon Orientation -- Cerenaut Part
The 5th WBA Hackathon Orientation  -- Cerenaut PartThe 5th WBA Hackathon Orientation  -- Cerenaut Part
The 5th WBA Hackathon Orientation -- Cerenaut Part
The Whole Brain Architecture Initiative
 
Image classification with Deep Neural Networks
Image classification with Deep Neural NetworksImage classification with Deep Neural Networks
Image classification with Deep Neural Networks
Yogendra Tamang
 
Convolution Neural Network (CNN)
Convolution Neural Network (CNN)Convolution Neural Network (CNN)
Convolution Neural Network (CNN)
Basit Rafiq
 
Visualizaing and understanding convolutional networks
Visualizaing and understanding convolutional networksVisualizaing and understanding convolutional networks
Visualizaing and understanding convolutional networks
SungminYou
 
Project on collision avoidance in static and dynamic environment
Project on collision avoidance in static and dynamic environmentProject on collision avoidance in static and dynamic environment
Project on collision avoidance in static and dynamic environment
gopaljee1989
 
A Survey about Object Retrieval
A Survey about Object RetrievalA Survey about Object Retrieval
A Survey about Object Retrieval
Nguyen Tuan
 
Convolutional neural network
Convolutional neural networkConvolutional neural network
Convolutional neural network
Itachi SK
 
Visualizing and Understanding Convolutional Networks
Visualizing and Understanding Convolutional NetworksVisualizing and Understanding Convolutional Networks
Visualizing and Understanding Convolutional Networks
Willy Marroquin (WillyDevNET)
 
A survey based on eeg classification
A survey based on eeg classificationA survey based on eeg classification
A survey based on eeg classification
ijitjournal
 
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
The Whole Brain Architecture Initiative
 

What's hot (20)

ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSIONADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
ADOPTING AND IMPLEMENTATION OF SELF ORGANIZING FEATURE MAP FOR IMAGE FUSION
 
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
TERRIAN IDENTIFICATION USING CO-CLUSTERED MODEL OF THE SWARM INTELLEGENCE & S...
 
Microscopy images segmentation algorithm based on shearlet neural network
Microscopy images segmentation algorithm based on shearlet neural networkMicroscopy images segmentation algorithm based on shearlet neural network
Microscopy images segmentation algorithm based on shearlet neural network
 
Efficient Neural Network Architecture for Image Classfication
Efficient Neural Network Architecture for Image ClassficationEfficient Neural Network Architecture for Image Classfication
Efficient Neural Network Architecture for Image Classfication
 
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
HUMAN ACTION RECOGNITION IN VIDEOS USING STABLE FEATURES
 
automated perimeters
automated perimetersautomated perimeters
automated perimeters
 
Image recognition
Image recognitionImage recognition
Image recognition
 
Ag044216224
Ag044216224Ag044216224
Ag044216224
 
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
CONVOLUTIONAL NEURAL NETWORK BASED FEATURE EXTRACTION FOR IRIS RECOGNITION
 
Random Neural Network (Erol) by Engr. Edgar Carrillo II
Random Neural Network (Erol) by Engr. Edgar Carrillo IIRandom Neural Network (Erol) by Engr. Edgar Carrillo II
Random Neural Network (Erol) by Engr. Edgar Carrillo II
 
The 5th WBA Hackathon Orientation -- Cerenaut Part
The 5th WBA Hackathon Orientation  -- Cerenaut PartThe 5th WBA Hackathon Orientation  -- Cerenaut Part
The 5th WBA Hackathon Orientation -- Cerenaut Part
 
Image classification with Deep Neural Networks
Image classification with Deep Neural NetworksImage classification with Deep Neural Networks
Image classification with Deep Neural Networks
 
Convolution Neural Network (CNN)
Convolution Neural Network (CNN)Convolution Neural Network (CNN)
Convolution Neural Network (CNN)
 
Visualizaing and understanding convolutional networks
Visualizaing and understanding convolutional networksVisualizaing and understanding convolutional networks
Visualizaing and understanding convolutional networks
 
Project on collision avoidance in static and dynamic environment
Project on collision avoidance in static and dynamic environmentProject on collision avoidance in static and dynamic environment
Project on collision avoidance in static and dynamic environment
 
A Survey about Object Retrieval
A Survey about Object RetrievalA Survey about Object Retrieval
A Survey about Object Retrieval
 
Convolutional neural network
Convolutional neural networkConvolutional neural network
Convolutional neural network
 
Visualizing and Understanding Convolutional Networks
Visualizing and Understanding Convolutional NetworksVisualizing and Understanding Convolutional Networks
Visualizing and Understanding Convolutional Networks
 
A survey based on eeg classification
A survey based on eeg classificationA survey based on eeg classification
A survey based on eeg classification
 
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
脳とAIの接点から何を学びうるのか@第5回WBAシンポジウム: 銅谷賢治
 

Similar to Learning Anticipation via Spiking Networks: Application to Navigation Control

Unsupervised representation learning for gaze estimation
Unsupervised representation learning for gaze estimationUnsupervised representation learning for gaze estimation
Unsupervised representation learning for gaze estimation
Jaey Jeong
 
Design and Development of Intelligent Navigation Control Systems for Autonomo...
Design and Development of Intelligent Navigation Control Systems for Autonomo...Design and Development of Intelligent Navigation Control Systems for Autonomo...
Design and Development of Intelligent Navigation Control Systems for Autonomo...
IJEID :: International Journal of Excellence Innovation and Development
 
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
sugiuralab
 
Deep Learning in Recommender Systems - RecSys Summer School 2017
Deep Learning in Recommender Systems - RecSys Summer School 2017Deep Learning in Recommender Systems - RecSys Summer School 2017
Deep Learning in Recommender Systems - RecSys Summer School 2017
Balázs Hidasi
 
Human action recognition with kinect using a joint motion descriptor
Human action recognition with kinect using a joint motion descriptorHuman action recognition with kinect using a joint motion descriptor
Human action recognition with kinect using a joint motion descriptor
Soma Boubou
 
Artificial Neural Network based Mobile Robot Navigation
Artificial Neural Network based Mobile Robot NavigationArtificial Neural Network based Mobile Robot Navigation
Artificial Neural Network based Mobile Robot Navigation
Mithun Chowdhury
 
Pedestrian dead reckoning indoor localization based on os-elm
Pedestrian dead reckoning indoor localization based on os-elmPedestrian dead reckoning indoor localization based on os-elm
Pedestrian dead reckoning indoor localization based on os-elm
Alwin Poulose
 
Robot navigation in unknown environment with obstacle recognition using laser...
Robot navigation in unknown environment with obstacle recognition using laser...Robot navigation in unknown environment with obstacle recognition using laser...
Robot navigation in unknown environment with obstacle recognition using laser...
IJECEIAES
 
Smart Room Gesture Control
Smart Room Gesture ControlSmart Room Gesture Control
Smart Room Gesture Control
Giwrgos Paraskevopoulos
 
Computational Tools for Extracting, Representing and Analyzing Facial Features
Computational Tools for Extracting, Representing and Analyzing Facial FeaturesComputational Tools for Extracting, Representing and Analyzing Facial Features
Computational Tools for Extracting, Representing and Analyzing Facial Features
saulnml
 
Lecture 4 neural networks
Lecture 4 neural networksLecture 4 neural networks
Lecture 4 neural networks
ParveenMalik18
 
Bci
BciBci
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
Masaya Kaneko
 
Bci
BciBci
khelchandra project on ai
khelchandra project on aikhelchandra project on ai
khelchandra project on ai
gopaljee1989
 
Henrik Christensen - Vision for Co-robot Applications
Henrik Christensen - Vision for Co-robot ApplicationsHenrik Christensen - Vision for Co-robot Applications
Henrik Christensen - Vision for Co-robot Applications
Daniel Huber
 
Henrik Christensen - Vision for co-robot applications
Henrik Christensen  -  Vision for co-robot applicationsHenrik Christensen  -  Vision for co-robot applications
Henrik Christensen - Vision for co-robot applications
Daniel Huber
 
Lecture 06: Features and Uncertainty
Lecture 06: Features and UncertaintyLecture 06: Features and Uncertainty
Lecture 06: Features and Uncertainty
University of Colorado at Boulder
 
Communityday2013
Communityday2013Communityday2013
Communityday2013
Matteo Valoriani
 
Lucio marcenaro tue summer_school
Lucio marcenaro tue summer_schoolLucio marcenaro tue summer_school
Lucio marcenaro tue summer_school
Jun Hu
 

Similar to Learning Anticipation via Spiking Networks: Application to Navigation Control (20)

Unsupervised representation learning for gaze estimation
Unsupervised representation learning for gaze estimationUnsupervised representation learning for gaze estimation
Unsupervised representation learning for gaze estimation
 
Design and Development of Intelligent Navigation Control Systems for Autonomo...
Design and Development of Intelligent Navigation Control Systems for Autonomo...Design and Development of Intelligent Navigation Control Systems for Autonomo...
Design and Development of Intelligent Navigation Control Systems for Autonomo...
 
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
Wearable Accelerometer Optimal Positions for Human Motion Recognition(LifeTec...
 
Deep Learning in Recommender Systems - RecSys Summer School 2017
Deep Learning in Recommender Systems - RecSys Summer School 2017Deep Learning in Recommender Systems - RecSys Summer School 2017
Deep Learning in Recommender Systems - RecSys Summer School 2017
 
Human action recognition with kinect using a joint motion descriptor
Human action recognition with kinect using a joint motion descriptorHuman action recognition with kinect using a joint motion descriptor
Human action recognition with kinect using a joint motion descriptor
 
Artificial Neural Network based Mobile Robot Navigation
Artificial Neural Network based Mobile Robot NavigationArtificial Neural Network based Mobile Robot Navigation
Artificial Neural Network based Mobile Robot Navigation
 
Pedestrian dead reckoning indoor localization based on os-elm
Pedestrian dead reckoning indoor localization based on os-elmPedestrian dead reckoning indoor localization based on os-elm
Pedestrian dead reckoning indoor localization based on os-elm
 
Robot navigation in unknown environment with obstacle recognition using laser...
Robot navigation in unknown environment with obstacle recognition using laser...Robot navigation in unknown environment with obstacle recognition using laser...
Robot navigation in unknown environment with obstacle recognition using laser...
 
Smart Room Gesture Control
Smart Room Gesture ControlSmart Room Gesture Control
Smart Room Gesture Control
 
Computational Tools for Extracting, Representing and Analyzing Facial Features
Computational Tools for Extracting, Representing and Analyzing Facial FeaturesComputational Tools for Extracting, Representing and Analyzing Facial Features
Computational Tools for Extracting, Representing and Analyzing Facial Features
 
Lecture 4 neural networks
Lecture 4 neural networksLecture 4 neural networks
Lecture 4 neural networks
 
Bci
BciBci
Bci
 
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
論文読み会@AIST (Deep Virtual Stereo Odometry [ECCV2018])
 
Bci
BciBci
Bci
 
khelchandra project on ai
khelchandra project on aikhelchandra project on ai
khelchandra project on ai
 
Henrik Christensen - Vision for Co-robot Applications
Henrik Christensen - Vision for Co-robot ApplicationsHenrik Christensen - Vision for Co-robot Applications
Henrik Christensen - Vision for Co-robot Applications
 
Henrik Christensen - Vision for co-robot applications
Henrik Christensen  -  Vision for co-robot applicationsHenrik Christensen  -  Vision for co-robot applications
Henrik Christensen - Vision for co-robot applications
 
Lecture 06: Features and Uncertainty
Lecture 06: Features and UncertaintyLecture 06: Features and Uncertainty
Lecture 06: Features and Uncertainty
 
Communityday2013
Communityday2013Communityday2013
Communityday2013
 
Lucio marcenaro tue summer_school
Lucio marcenaro tue summer_schoolLucio marcenaro tue summer_school
Lucio marcenaro tue summer_school
 

More from Seonghyun Kim

코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
Seonghyun Kim
 
뇌의 정보처리와 멀티모달 인공지능
뇌의 정보처리와 멀티모달 인공지능뇌의 정보처리와 멀티모달 인공지능
뇌의 정보처리와 멀티모달 인공지능
Seonghyun Kim
 
인공지능과 윤리
인공지능과 윤리인공지능과 윤리
인공지능과 윤리
Seonghyun Kim
 
한국어 개체명 인식 과제에서의 의미 모호성 연구
한국어 개체명 인식 과제에서의 의미 모호성 연구한국어 개체명 인식 과제에서의 의미 모호성 연구
한국어 개체명 인식 과제에서의 의미 모호성 연구
Seonghyun Kim
 
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
Seonghyun Kim
 
Backpropagation and the brain review
Backpropagation and the brain reviewBackpropagation and the brain review
Backpropagation and the brain review
Seonghyun Kim
 
Theories of error back propagation in the brain review
Theories of error back propagation in the brain reviewTheories of error back propagation in the brain review
Theories of error back propagation in the brain review
Seonghyun Kim
 
KorQuAD v1.0 참관기
KorQuAD v1.0 참관기KorQuAD v1.0 참관기
KorQuAD v1.0 참관기
Seonghyun Kim
 
딥러닝 기반 자연어 언어모델 BERT
딥러닝 기반 자연어 언어모델 BERT딥러닝 기반 자연어 언어모델 BERT
딥러닝 기반 자연어 언어모델 BERT
Seonghyun Kim
 
Enriching Word Vectors with Subword Information
Enriching Word Vectors with Subword InformationEnriching Word Vectors with Subword Information
Enriching Word Vectors with Subword Information
Seonghyun Kim
 
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
BERT: Pre-training of Deep Bidirectional Transformers for Language UnderstandingBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Seonghyun Kim
 
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
Seonghyun Kim
 
챗봇의 역사
챗봇의 역사챗봇의 역사
챗봇의 역사
Seonghyun Kim
 

More from Seonghyun Kim (13)

코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
코드 스위칭 코퍼스 기반 다국어 LLM의 지식 전이 연구
 
뇌의 정보처리와 멀티모달 인공지능
뇌의 정보처리와 멀티모달 인공지능뇌의 정보처리와 멀티모달 인공지능
뇌의 정보처리와 멀티모달 인공지능
 
인공지능과 윤리
인공지능과 윤리인공지능과 윤리
인공지능과 윤리
 
한국어 개체명 인식 과제에서의 의미 모호성 연구
한국어 개체명 인식 과제에서의 의미 모호성 연구한국어 개체명 인식 과제에서의 의미 모호성 연구
한국어 개체명 인식 과제에서의 의미 모호성 연구
 
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
파이콘 한국 2020) 파이썬으로 구현하는 신경세포 기반의 인공 뇌 시뮬레이터
 
Backpropagation and the brain review
Backpropagation and the brain reviewBackpropagation and the brain review
Backpropagation and the brain review
 
Theories of error back propagation in the brain review
Theories of error back propagation in the brain reviewTheories of error back propagation in the brain review
Theories of error back propagation in the brain review
 
KorQuAD v1.0 참관기
KorQuAD v1.0 참관기KorQuAD v1.0 참관기
KorQuAD v1.0 참관기
 
딥러닝 기반 자연어 언어모델 BERT
딥러닝 기반 자연어 언어모델 BERT딥러닝 기반 자연어 언어모델 BERT
딥러닝 기반 자연어 언어모델 BERT
 
Enriching Word Vectors with Subword Information
Enriching Word Vectors with Subword InformationEnriching Word Vectors with Subword Information
Enriching Word Vectors with Subword Information
 
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
BERT: Pre-training of Deep Bidirectional Transformers for Language UnderstandingBERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
 
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
Korean-optimized Word Representations for Out of Vocabulary Problems caused b...
 
챗봇의 역사
챗봇의 역사챗봇의 역사
챗봇의 역사
 

Recently uploaded

SAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdfSAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdf
KrushnaDarade1
 
DMARDs Pharmacolgy Pharm D 5th Semester.pdf
DMARDs Pharmacolgy Pharm D 5th Semester.pdfDMARDs Pharmacolgy Pharm D 5th Semester.pdf
DMARDs Pharmacolgy Pharm D 5th Semester.pdf
fafyfskhan251kmf
 
NuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyerNuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyer
pablovgd
 
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdfTopic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
TinyAnderson
 
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốtmô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
HongcNguyn6
 
The debris of the ‘last major merger’ is dynamically young
The debris of the ‘last major merger’ is dynamically youngThe debris of the ‘last major merger’ is dynamically young
The debris of the ‘last major merger’ is dynamically young
Sérgio Sacani
 
20240520 Planning a Circuit Simulator in JavaScript.pptx
20240520 Planning a Circuit Simulator in JavaScript.pptx20240520 Planning a Circuit Simulator in JavaScript.pptx
20240520 Planning a Circuit Simulator in JavaScript.pptx
Sharon Liu
 
Shallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptxShallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptx
Gokturk Mehmet Dilci
 
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
University of Maribor
 
Medical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptxMedical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptx
terusbelajar5
 
Deep Software Variability and Frictionless Reproducibility
Deep Software Variability and Frictionless ReproducibilityDeep Software Variability and Frictionless Reproducibility
Deep Software Variability and Frictionless Reproducibility
University of Rennes, INSA Rennes, Inria/IRISA, CNRS
 
Oedema_types_causes_pathophysiology.pptx
Oedema_types_causes_pathophysiology.pptxOedema_types_causes_pathophysiology.pptx
Oedema_types_causes_pathophysiology.pptx
muralinath2
 
Chapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisisChapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisis
tonzsalvador2222
 
Orion Air Quality Monitoring Systems - CWS
Orion Air Quality Monitoring Systems - CWSOrion Air Quality Monitoring Systems - CWS
Orion Air Quality Monitoring Systems - CWS
Columbia Weather Systems
 
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdfUnveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Erdal Coalmaker
 
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
Ana Luísa Pinho
 
Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.
Nistarini College, Purulia (W.B) India
 
The binding of cosmological structures by massless topological defects
The binding of cosmological structures by massless topological defectsThe binding of cosmological structures by massless topological defects
The binding of cosmological structures by massless topological defects
Sérgio Sacani
 
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptxThe use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
MAGOTI ERNEST
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
moosaasad1975
 

Recently uploaded (20)

SAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdfSAR of Medicinal Chemistry 1st by dk.pdf
SAR of Medicinal Chemistry 1st by dk.pdf
 
DMARDs Pharmacolgy Pharm D 5th Semester.pdf
DMARDs Pharmacolgy Pharm D 5th Semester.pdfDMARDs Pharmacolgy Pharm D 5th Semester.pdf
DMARDs Pharmacolgy Pharm D 5th Semester.pdf
 
NuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyerNuGOweek 2024 Ghent programme overview flyer
NuGOweek 2024 Ghent programme overview flyer
 
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdfTopic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
Topic: SICKLE CELL DISEASE IN CHILDREN-3.pdf
 
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốtmô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
mô tả các thí nghiệm về đánh giá tác động dòng khí hóa sau đốt
 
The debris of the ‘last major merger’ is dynamically young
The debris of the ‘last major merger’ is dynamically youngThe debris of the ‘last major merger’ is dynamically young
The debris of the ‘last major merger’ is dynamically young
 
20240520 Planning a Circuit Simulator in JavaScript.pptx
20240520 Planning a Circuit Simulator in JavaScript.pptx20240520 Planning a Circuit Simulator in JavaScript.pptx
20240520 Planning a Circuit Simulator in JavaScript.pptx
 
Shallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptxShallowest Oil Discovery of Turkiye.pptx
Shallowest Oil Discovery of Turkiye.pptx
 
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
Remote Sensing and Computational, Evolutionary, Supercomputing, and Intellige...
 
Medical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptxMedical Orthopedic PowerPoint Templates.pptx
Medical Orthopedic PowerPoint Templates.pptx
 
Deep Software Variability and Frictionless Reproducibility
Deep Software Variability and Frictionless ReproducibilityDeep Software Variability and Frictionless Reproducibility
Deep Software Variability and Frictionless Reproducibility
 
Oedema_types_causes_pathophysiology.pptx
Oedema_types_causes_pathophysiology.pptxOedema_types_causes_pathophysiology.pptx
Oedema_types_causes_pathophysiology.pptx
 
Chapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisisChapter 12 - climate change and the energy crisis
Chapter 12 - climate change and the energy crisis
 
Orion Air Quality Monitoring Systems - CWS
Orion Air Quality Monitoring Systems - CWSOrion Air Quality Monitoring Systems - CWS
Orion Air Quality Monitoring Systems - CWS
 
Unveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdfUnveiling the Energy Potential of Marshmallow Deposits.pdf
Unveiling the Energy Potential of Marshmallow Deposits.pdf
 
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
Deep Behavioral Phenotyping in Systems Neuroscience for Functional Atlasing a...
 
Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.Nucleic Acid-its structural and functional complexity.
Nucleic Acid-its structural and functional complexity.
 
The binding of cosmological structures by massless topological defects
The binding of cosmological structures by massless topological defectsThe binding of cosmological structures by massless topological defects
The binding of cosmological structures by massless topological defects
 
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptxThe use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
The use of Nauplii and metanauplii artemia in aquaculture (brine shrimp).pptx
 
What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.What is greenhouse gasses and how many gasses are there to affect the Earth.
What is greenhouse gasses and how many gasses are there to affect the Earth.
 

Learning Anticipation via Spiking Networks: Application to Navigation Control

  • 1. Learning Anticipation via Spiking Networks: Application to Navigation Control Paolo Arena et al. 2009 IEEE Transactions on Neural Network
  • 3. Spatial Information Processing ▪ Rat has the navigation cells that represent the place[1], grid[2], boundary[3] and direction[4] information in the environment [1] O’keefe and Dostrovsky, 1971, Brain Res [3] Lever et al., 2009, J Neurosci. [5] Moser et al., 2008, Annu Rev Neurosci [2] Fyhn et al., 2004, Science [4] Taube et al., 1990, J Neurosci. [6] Zhang et al., 2013, Philos Trans R Soc Lond B Biol Sci. Place cell [5] Head direction ( °) Firingrate(Hz) Preferred direction Head direction cell [4] Boundary vector cell [7] Grid cell [5] Rat trajectory Cell spikes
  • 4. Neural Network Architecture ▪ Artificial neural network ▫ Output is calculated by summation of inputs with weight vectors ▫ The weight vector is modulated to get the right output ▪ Spiking neural network ▫ Output is the result of neural activity by presynaptic inputs ▫ The synaptic weight is modulated by biological learning rule
  • 5. Spike-timing Dependent Plasticity (STDP) ▪ Excitatory postsynaptic current (EPSC) depends on the relative timing of pre- and postsynaptic action potentials ▪ Unsupervised learning rule [7] Bi and Poo, 1998, J Neurosci. Δt (tpost – tpre) ChangeinEPSCamplitude(%) Pre Post [7]
  • 6. Learning Factors ▪ Classical conditioning provided the source of inspiration for the implementation of navigation task algorithm Unconditioned stimuli Unconditioned response Conditioned stimuli Conditioned response
  • 7. Method ▪ The Spiking Network Model ▫ Sensory neurons are neurons connected to sensors ▫ Motor neurons drive robot motors ▫ Interneurons ▫ Each single neurons are modeled by Izhikevich model v: membrane potential u: recovery variable I: input current [8] Izhikevich, IEEE, 2004 20 ms a = 0.02 b = -0.1 c = -55 d = 6 i j [8]
  • 8. Method ▪ STDP modification function Δt (tpre – tpost) ഥ𝒈 𝒎𝒂𝒙 𝜏+ = 20 𝜏− = 10 𝐴+ = 0.02 𝐴− = −0.02 [9] Song et al., 2000, Nat Neurosci. [9]
  • 9. Method ▪ Robot simulation ▫ Dual-drive wheeled virtual robot (right motor: RM, left motor: LM) ▫ Two collision sensor (US) ▫ Two rand finder sensor ([0°, -45°], [0°, 45°], CS) ▫ Two proximity sensor (for target) Left range finder sensor Right range finder sensor Range of proximity sensor Camera view (black line)
  • 10. Method ▪ Robot experiment ▫ Real roving robot ▫ Lynxmotion 4WD2 Robot ▫ Four infrared distance sensors (14 to 80 cm) ▫ Collision (threshold at 23 cm) ▫ RF04 USB radio telemetry module
  • 11. Method ▪ Network for obstacle avoidance ▫ The robot moves (0.3 r.u. for each simulation step, time window: 300 ms) in an environment filled with randomly placed obstacles Collision sensor
  • 12. Method Proximity target sensor ▪ The network for target approaching ▫ The barycenter of red object is provided to the network of nine neurons as a visual input
  • 13. Method ▪ The network for navigation with visual cue
  • 14. Result – Obstacle Avoidance ▪ Obstacle avoidance Robot is driven by UR Robot is driven by CR Robot is driven by CR Obstacleavoidancenumber
  • 15. Result - Obstacle Avoidance ▪ The number of collision and wrong decision is decreased by STDP learning rule NumberofcollisionDistancebetweenrobotand closestobstacles
  • 16. Result – Synaptic Weight Change ▪ Obstacle avoidance Potentiation Potentiation Depression Depression
  • 17. Result – Obstacle Avoidance ▪ Obstacle avoidance ▫ Trajectories obtained in the case in which the robot faces one obstacle Before learning End of learning phase
  • 18. Result – Synaptic Weight Change ▪ Obstacle avoidance ▫ Robot experiments
  • 19. Result – Target Approaching ▪ Target approaching Without learning After learning (without visual input) (with visual input) Long run simulation Range of proximity sensor
  • 20. Result – Target Approaching (without visual input) (with visual input) Before learning End of learning phase Two targets
  • 21. Result - Navigation with Visual Cue Cases of training environments Markers Case (a) Solving maze based on case (a)
  • 22. Conclusion ▪ Biological spiking neural network shows the efficiency for the dynamic systems instead of a static relationship between input and output ▪ STDP unsupervised learning rule is one of a simple way for navigation algorithm ▪ The spiking cells are suitable for investigate the temporal code of neural activity
  • 23. Discussion ▪ The design of neural network is not considered for neuromorphic ▪ These networks cannot recall the specific spatial memory (place)