2017 Tutorial - Deep Learning for Dialogue Systems

Deep Learning for Dialogue Systemsdeepdialogue.miulab.tw
YUN-NUNG (VIVIAN) CHEN ASLI CELIKYILMAZ DILEK HAKKANI-TÜR

2
Outline
 Introduction
 Background Knowledge
 Neural Network Basics
 Reinforcement Learning
 Modular Dialogue System
 Spoken/Natural Language Understanding (SLU/NLU)
 Dialogue Management
 Dialogue State Tracking (DST)
 Dialogue Policy Optimization
 Natural Language Generation (NLG)
 Evaluation
 Recent Trends and Challenges
 End-to-End Neural Dialogue System
 Multimodality
 Dialogue Breath
 Dialogue Depth
2
Material: http://deepdialogue.miulab.tw
Break

4
Early 1990s
Early 2000s
2017
Multi-modal systems
e.g., Microsoft MiPad, Pocket PC
Keyword Spotting
(e.g., AT&T)
System: “Please say collect,
calling card, person, third
number, or operator”
TV Voice Search
e.g., Bing on Xbox
Intent Determination
(Nuance’s Emily™, AT&T HMIHY)
User: “Uh…we want to move…we
want to change our phone line
from this house to another house”
Task-specific argument extraction
(e.g., Nuance, SpeechWorks)
User: “I want to fly from Boston
to New York next week.”
Brief History of Dialogue Systems
Apple Siri
(2011)
Google Now (2012)
Facebook M & Bot
(2015)
Google Home
(2016)
Microsoft Cortana
(2014)
Amazon Alexa/Echo
(2014)
Google Assistant
(2016)
DARPA
CALO Project
Virtual Personal Assistants

5
Language Empowering Intelligent Assistant
Apple Siri (2011) Google Now (2012)
Facebook M & Bot (2015) Google Home (2016)
Microsoft Cortana (2014)
Amazon Alexa/Echo (2014)
Google Assistant (2016)
Apple HomePod (2017)

6
Why We Need?
 Get things done
 E.g. set up alarm/reminder, take note
 Easy access to structured data, services and apps
 E.g. find docs/photos/restaurants
 Assist your daily schedule and routine
 E.g. commute alerts to/from work
 Be more productive in managing your work and personal life
6

7
Why Natural Language?
 Global Digital Statistics (2015 January)
7
Global Population
7.21B
Active Internet Users
3.01B
Active Social
Media Accounts
2.08B
Active Unique
Mobile Users
3.65B
The more natural and convenient input of devices evolves towards speech.

8
Spoken Dialogue System (SDS)
 Spoken dialogue systems are intelligent agents that are able to help users finish tasks more
efficiently via spoken interactions.
 Spoken dialogue systems are being incorporated into various devices (smart-phones, smart TVs, in-
car navigating system, etc).
8
JARVIS – Iron Man’s Personal Assistant Baymax – Personal Healthcare Companion
Good dialogue systems assist users to access information conveniently and finish tasks efficiently.

9
App  Bot
 A bot is responsible for a “single” domain, similar to an app
9
Users can initiate dialogues instead of following the GUI design

10
GUI v.s. CUI (Conversational UI)
10
https://github.com/enginebai/Movie-lol-android

11
GUI v.s. CUI (Conversational UI)
Website/APP’s GUI Msg’s CUI
Situation Navigation, no specific goal Searching, with specific goal
Information Quantity More Less
Information Precision Low High
Display Structured Non-structured
Interface Graphics Language
Manipulation Click mainly use texts or speech as input
Learning Need time to learn and adapt No need to learn
Entrance App download Incorporated in any msg-based interface
Flexibility Low, like machine manipulation High, like converse with a human
11

12
Challenges
 Variability in Natural Language
 Robustness
 Recall/Precision Trade-off
 Meaning Representation
 Common Sense, World Knowledge
 Ability to Learn
 Transparency
12

Two Branches of Bots
 Personal assistant, helps users achieve a certain task
 Combination of rules and statistical components
 POMDP for spoken dialog systems (Williams and Young, 2007)
 End-to-end trainable task-oriented dialogue system (Wen et al.,
2016)
 End-to-end reinforcement learning dialogue system (Li et al.,
2017; Zhao and Eskenazi, 2016)
 No specific goal, focus on natural responses
 Using variants of seq2seq model
 A neural conversation model (Vinyals and Le, 2015)
 Reinforcement learning for dialogue generation (Li et
al., 2016)
 Conversational contextual cues for response ranking
(AI-Rfou et al., 2016)
13
Task-Oriented Bot Chit-Chat Bot

14
Task-Oriented Dialogue System (Young, 2000)
14
Speech
Recognition
Language Understanding (LU)
• Domain Identification
• User Intent Detection
• Slot Filling
Dialogue Management (DM)
• Dialogue State Tracking (DST)
• Dialogue Policy
Natural Language
Generation (NLG)
Hypothesis
are there any action movies to
see this weekend
Semantic Frame
request_movie
genre=action, date=this weekend
System Action/Policy
request_location
Text response
Where are you located?
Text Input
Are there any action movies to see this weekend?
Speech Signal
Backend Action /
Knowledge Providers
http://rsta.royalsocietypublishing.org/content/358/1769/1389.short

15
Interaction Example
15
User
Intelligent
Agent Q: How does a dialogue system process this request?
Good Taiwanese eating places include Din Tai
Fung, Boiling Point, etc. What do you want to
choose? I can help you go there.
find a good eating place for taiwanese food

16
16
Speech
Recognition
• Slot Filling
• Dialogue Policy
Natural Language
Generation (NLG)
Hypothesis
see this weekend
Semantic Frame
request_movie
request_location
Text response
Text Input
Speech Signal
Backend Action /
Knowledge Providers

17
1. Domain Identification
Requires Predefined Domain Ontology
17
User
Organized Domain Knowledge (Database)Intelligent
Agent
Restaurant DB Taxi DB Movie DB
Classification!

18
2. Intent Detection
Requires Predefined Schema
18
User
Intelligent
Agent
Restaurant DB
FIND_RESTAURANT
FIND_PRICE
FIND_TYPE
:
Classification!

19
3. Slot Filling
Requires Predefined Schema
User
Intelligent
Agent
19
Restaurant DB
Restaurant Rating Type
Rest 1 good Taiwanese
Rest 2 bad Thai
: : :
FIND_RESTAURANT
rating=“good”
type=“taiwanese”
SELECT restaurant {
rest.rating=“good”
rest.type=“taiwanese”
}Semantic Frame Sequence Labeling
O O B-rating O O O B-type O

20
20
Speech
Recognition
• Slot Filling
• Dialogue Policy
Natural Language
Generation (NLG)
Hypothesis
see this weekend
Semantic Frame
request_movie
request_location
Text response
Text Input
Speech Signal
Backend Action /
Knowledge Providers

21
State Tracking
Requires Hand-Crafted States
User
Intelligent
Agent
21
location rating type
loc, rating
rating,
type
loc,
type
all
i want it near to my office
NULL

22
State Tracking
Requires Hand-Crafted States
User
Intelligent
Agent
22
loc, rating
rating,
type
loc,
type
all
i want it near to my office
NULL

23
State Tracking
Handling Errors and Confidence
User
Intelligent
Agent
find a good eating place for taixxxx food
23
FIND_RESTAURANT
rating=“good”
FIND_RESTAURANT
rating=“good”
type=“thai”
FIND_RESTAURANT
rating=“good”
loc, rating
rating,
type
loc,
type
all
NULL
?
?
rating=“good”,
type=“thai”
rating=“good”,
?
?

24
Dialogue Policy for Agent Action
 Inform(location=“Taipei 101”)
 “The nearest one is at Taipei 101”
 Request(location)
 “Where is your home?”
 Confirm(type=“taiwanese”)
 “Did you want Taiwanese food?”
24

25
Speech
Recognition
• Slot Filling
Hypothesis
see this weekend
Semantic Frame
request_movie
request_location
Text Input
Speech Signal
• Dialogue Policy
Backend Action /
Knowledge Providers
Natural Language
Generation (NLG)Text response

26
Output / Natural Language Generation
 Goal: generate natural language or GUI given the selected dialogue action for interactions
 Inform(location=“Taipei 101”)
 “The nearest one is at Taipei 101” v.s.
 Request(location)
 “Where is your home?” v.s.
 Confirm(type=“taiwanese”)
 “Did you want Taiwanese food?” v.s.
26

Background Knowledge27
Neural Network Basics
Reinforcement Learning

28
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
28

29
Machine Learning ≈ Looking for a Function
 Speech Recognition
 Image Recognition
 Go Playing
 Chat Bot
 f
 f
 f
 f
cat
“你好 (Hello) ”
5-5 (next move)
“Where is Westin?” “The address is…”
Given a large amount of data, the machine learns what the function f should be.

30
Machine Learning
30
Machine
Learning
Unsupervised
Learning
Supervised
Learning
Reinforcement
Learning
Deep learning is a type of machine learning approaches, called “neural networks”.

31
A Single Neuron
z
1w
2w
Nw
…1x
2x
Nx

b
 z
 z
zbias
y
  z
e
z 


1
1

Sigmoid function
Activation function
1
w, b are the parameters of this neuron
31

32
A Single Neuron
z
1w
2w
Nw…
1x
2x
Nx

b
bias
y
1





5.0"2"
5.0"2"
ynot
yis
A single neuron can only handle binary classification
32
MN
RRf :

33
A Layer of Neurons
 Handwriting digit classification
MN
RRf :
A layer of neurons can handle multiple possible output,
and the result depends on the max one
…
1x
2x
Nx

1
 1y

…
…
“1” or not
“2” or not
“3” or not
2y
3y
10 neurons/10 classes
Which
one is
max?

34
Deep Neural Networks (DNN)
 Fully connected feedforward network
1x
2x
……
Layer 1
…… 1y
2y
……
Layer 2
……
Layer L
……
……
……
Input Output
MyNx
vector
x
vector
y
Deep NN: multiple hidden layers
MN
RRf :

35
Recurrent Neural Network (RNN)
http://www.wildml.com/2015/09/recurrent-neural-networks-tutorial-part-1-introduction-to-rnns/
: tanh, ReLU
time
RNN can learn accumulated sequential information (time-series)

36
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
36

37
Reinforcement Learning
 RL is a general purpose framework for decision making
 RL is for an agent with the capacity to act
 Each action influences the agent’s future state
 Success is measured by a scalar reward signal
 Goal: select actions to maximize future reward

38
Scenario of Reinforcement Learning
Agent learns to take actions to maximize expected reward.
Environment
Observation ot Action at
Reward rt
If win, reward = 1
If loss, reward = -1
Otherwise, reward = 0
Next Move

39
Supervised v.s. Reinforcement
 Supervised
 Reinforcement
39
Hello ☺
Agent
……
Agent
……. ……. ……
Bad
“Hello” Say “Hi”
“Bye bye” Say “Good bye”
Learning from teacher
Learning from critics

40
Sequential Decision Making
 Goal: select actions to maximize total future reward
 Actions may have long-term consequences
 Reward may be delayed
 It may be better to sacrifice immediate reward to gain more long-term reward
40

41
Deep Reinforcement Learning
Environment
Observation Action
Reward
Function
Input
Function
Output
Used to pick the
best function
…
…
…
DNN

42
Reinforcing Learning
 Start from state s0
 Choose action a0
 Transit to s1 ~ P(s0, a0)
 Continue…
 Total reward:
Goal: select actions that maximize the expected total reward

43
Reinforcement Learning Approach
 Policy-based RL
 Search directly for optimal policy
 Value-based RL
 Estimate the optimal value function
 Model-based RL
 Build a model of the environment
 Plan (e.g. by lookahead) using model
is the policy achieving maximum future reward
is maximum value achievable under any policy

45
45
Speech
Recognition
• Slot Filling
• Dialogue Policy
Natural Language
Generation (NLG)
Hypothesis
see this weekend
Semantic Frame
request_movie
request_location
Text response
Text Input
Speech Signal
Backend Action /
Knowledge Providers
http://rsta.royalsocietypublishing.org/content/358/1769/1389.short

46
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
46

47
 Pipelined
47
1. Domain
Classification
2. Intent
Classification
3. Slot Filling

LU – Domain/Intent Classification
• Given a collection of utterances ui with labels ci, D= {(u1,c1),…,(un,cn)}
where ci ∊ C, train a model to estimate labels for new utterances uk.
Mainly viewed as an utterance classification task
48
find me a cheap taiwanese restaurant in oakland
Movies
Restaurants
Sports
Weather
Music
…
Find_movie
Buy_tickets
Find_restaurant
Book_table
Find_lyrics
…

49
DNN for Domain/Intent Classification – I (Sarikaya et al., 2011)
 Deep belief nets (DBN)
 Unsupervised training of weights
 Fine-tuning by back-propagation
 Compared to MaxEnt, SVM, and boosting
49
http://ieeexplore.ieee.org/abstract/document/5947649/

50
DNN for Domain/Intent Classification – II (Tur et al., 2012;
Deng et al., 2012)
 Deep convex networks (DCN)
 Simple classifiers are stacked to learn complex functions
 Feature selection of salient n-grams
 Extension to kernel-DCN
50
http://ieeexplore.ieee.org/abstract/document/6289054/; http://ieeexplore.ieee.org/abstract/document/6424224/

51
DNN for Domain/Intent Classification – III (Ravuri & Stolcke, 2015)
51
https://www.microsoft.com/en-us/research/wp-content/uploads/2016/02/RNNLM_addressee.pdf
Intent decision after reading all words performs better
 RNN and LSTMs for utterance classification

52
DNN for Dialogue Act Classification – IV (Lee & Dernoncourt, 2016)
52
 RNN and CNNs for dialogue act classification

LU – Slot Filling
53
flights from Boston to New York today
O O B-city O B-city I-city O
O O B-dept O B-arrival I-arrival B-date
As a sequence
tagging task
• Given a collection tagged word sequences, S={((w1,1,w1,2,…, w1,n1),
(t1,1,t1,2,…,t1,n1)), ((w2,1,w2,2,…,w2,n2), (t2,1,t2,2,…,t2,n2)) …}
where ti ∊ M, the goal is to estimate tags for a new word sequence.
flights from Boston to New York today
Entity Tag
Slot Tag

54
Recurrent Neural Nets for Slot Tagging – I (Yao et al, 2013;
Mesnil et al, 2015)
 Variations:
a. RNNs with LSTM cells
b. Input, sliding window of n-grams
c. Bi-directional LSTMs
𝑤0 𝑤1 𝑤2 𝑤 𝑛
ℎ0
𝑓
ℎ1
𝑓
ℎ2
𝑓
ℎ 𝑛
𝑓
ℎ0
𝑏
ℎ1
𝑏
ℎ2
𝑏 ℎ 𝑛
𝑏
𝑦0 𝑦1 𝑦2 𝑦𝑛
(b) LSTM-LA (c) bLSTM
ℎ0 ℎ1 ℎ2 ℎ 𝑛
(a) LSTM
http://131.107.65.14/en-us/um/people/gzweig/Pubs/Interspeech2013RNNLU.pdf; http://dl.acm.org/citation.cfm?id=2876380

55
Recurrent Neural Nets for Slot Tagging – II (Kurata et al., 2016;
Simonnet et al., 2015)
 Encoder-decoder networks
 Leverages sentence level information
 Attention-based encoder-decoder
 Use of attention (as in MT) in the
encoder-decoder network
 Attention is estimated using a feed-
forward network with input: ht and st at
time t
𝑤 𝑛 𝑤2 𝑤1 𝑤0
ℎ 𝑛 ℎ2 ℎ1 ℎ0
𝑠0 𝑠1 𝑠2 𝑠 𝑛
ci
ℎ0 ℎ 𝑛…
http://www.aclweb.org/anthology/D16-1223

56
Recurrent Neural Nets for Slot Tagging – III (Jaech et al., 2016;
Tafforeau et al., 2016)
 Multi-task learning
 Goal: exploit data from domains/tasks with a lot of data to improve ones
with less data
 Lower layers are shared across domains/tasks
 Output layer is specific to task
56
https://arxiv.org/abs/1604.00117; http://www.sensei-conversation.eu/wp-content/uploads/2016/11/favre_is2016b.pdf

57
Joint Segmentation and Slot Tagging (Zhai et al., 2017)
 Encoder that segments
 Decoder that tags the segments
57
https://arxiv.org/pdf/1701.04027.pdf

ht-
1
ht+
1
ht
W W W W
taiwanese
B-type
U
food
U
please
U
V
O
V
O
V
hT+1
EOS
U
FIND_REST
V
Slot Filling Intent Prediction
Joint Semantic Frame Parsing
Sequence-
based
(Hakkani-Tur
et al., 2016)
• Slot filling and
intent prediction
in the same
output sequence
Parallel
(Liu and
Lane, 2016)
• Intent prediction
and slot filling
are performed
in two branches
58 https://www.microsoft.com/en-us/research/wp-content/uploads/2016/06/IS16_MultiJoint.pdf; https://arxiv.org/abs/1609.01454

59
Contextual LU
59
just sent email to bob about fishing this weekend
O O O O
B-contact_name
O
B-subject I-subject I-subject
U
S
I send_emailD communication
 send_email(contact_name=“bob”, subject=“fishing this weekend”)
are we going to fish this weekend
U1
S2
 send_email(message=“are we going to fish this weekend”)
send email to bob
U2
 send_email(contact_name=“bob”)
B-message
I-message
I-message I-message I-message
I-message I-message
B-contact_nameS1
Domain Identification  Intent Prediction  Slot Filling

60
Contextual LU
 User utterances are highly ambiguous in isolation
Cascal, for 6.
#people time
?
Book a table for 10 people tonight.
Which restaurant would you like to book a table for?
Restaurant
Booking

61
Contextual LU (Bhargava et al., 2013; Hori et al, 2015)
 Leveraging contexts
 Used for individual tasks
 Seq2Seq model
 Words are input one at a time, tags are output at the end of each utterance
 Extension: LSTM with speaker role dependent layers
61
https://www.merl.com/publications/docs/TR2015-134.pdf

62
End-to-End Memory Networks (Sukhbaatar et al, 2015)
U: “i d like to purchase tickets to see deepwater horizon”
S: “for which theatre”
U: “angelika”
S: “you want them for angelika theatre?”
U: “yes angelika”
S: “how many tickets would you like ?”
U: “3 tickets for saturday”
S: “What time would you like ?”
U: “Any time on saturday is fine”
S: “okay , there is 4:10 pm , 5:40 pm and 9:20 pm”
U: “Let’s do 5:40”
m0
mi
mn-1
u

63
E2E MemNN for Contextual LU (Chen et al., 2016)
63
u
Knowledge Attention Distributionpi
mi
Memory Representation
Weighted
Sum
h
∑ Wkg
o
Knowledge Encoding
Representation
history utterances {xi}
current utterance
c
Inner
Product
Sentence
Encoder
RNNin
x1 x2 xi…
Contextual
Sentence Encoder
x1 x2 xi…
RNNmem
slot tagging sequence y
ht-1 ht
V V
W W W
wt-1 wt
yt-1 yt
U UM M
1. Sentence Encoding 2. Knowledge Attention 3. Knowledge Encoding
Idea: additionally incorporating contextual knowledge during slot tagging
 track dialogue states in a latent way
RNN Tagger
https://www.microsoft.com/en-us/research/wp-content/uploads/2016/06/IS16_ContextualSLU.pdf

64
Analysis of Attention
U: “i d like to purchase tickets to see deepwater horizon”
S: “for which theatre”
U: “angelika”
S: “you want them for angelika theatre?”
U: “yes angelika”
S: “how many tickets would you like ?”
U: “3 tickets for saturday”
S: “What time would you like ?”
U: “Any time on saturday is fine”
S: “okay , there is 4:10 pm , 5:40 pm and 9:20 pm”
U: “Let’s do 5:40”
0.69
0.13
0.16

65
Sequential Dialogue Encoder Network (Bapna et al., 2017)
 Past and current turn encodings input to a feed forward network
65
Bapna et.al., SIGDIAL 2017

66
Structural LU (Chen et al., 2016)
 K-SAN: prior knowledge as a teacher
66
Knowledge
Encoding
Sentence
Encoding
Inner
Product
mi
Knowledge Attention Distribution
pi
Encoded Knowledge Representation
Weighted
Sum
∑
Knowledge-
Guided
Representation
slot tagging sequence
knowledge-guided structure {xi}
showme theflights fromseattleto sanfrancisco
ROOT
Input Sentence
W W W W
wt-1
yt-1
U
Mwt
U
wt+1
U
V
yt
V
yt+1
V
MM
RNN Tagger
Knowledge Encoding Module
http://arxiv.org/abs/1609.03286

67
 Sentence structural knowledge stored as memory
67
Semantics (AMR Graph)
show
me
the
flights
from
seattle
to
san
francisco
ROOT
1.
3.
4.
2.
show
you
flight
I
1.
2.
4.
city
city
Seattle
San Francisco
3.
Sentence s show me the flights from seattle to san francisco
Syntax (Dependency Tree)

68
 Sentence structural knowledge stored as memory
Using less training data with K-SAN allows the model pay the similar attention
to the salient substructures that are important for tagging.

69
LU Importance (Li et al., 2017)
 Compare different types of LU errors
Slot filling is more important than intent detection in language understanding
Sensitivity to Intent Error Sensitivity to Slot Error

70
LU Evaluation
 Metrics
 Sub-sentence-level: intent accuracy, slot F1
 Sentence-level: whole frame accuracy
70

71
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
71

72
Elements of Dialogue Management
72(Figure from Gašić)
Dialogue State Tracking

73
Dialogue State Tracking (DST)
 Maintain a probabilistic distribution instead of a 1-best prediction for
better robustness
73
Incorrect
for both!

74
Dialogue State Tracking (DST)
 Maintain a probabilistic distribution instead of a 1-best prediction for
better robustness to SLU errors or ambiguous input
74
How can I help you?
Book a table at Sumiko for 5
How many people?
3
Slot Value
# people 5 (0.5)
time 5 (0.5)
Slot Value
# people 3 (0.8)
time 5 (0.8)

75
Multi-Domain Dialogue State Tracking (DST)
 A full representation of the system's belief of
the user's goal at any point during the dialogue
 Used for making API calls
75
Do you wanna take Angela
to go see a movie tonight?
Sure, I will be home by 6.
Let's grab dinner before the
movie.
How about some Mexican?
Let's go to Vive Sol and see
Inferno after that.
Angela wants to watch the
Trolls movie.
Ok. Lets catch the 8 pm
show.
Inferno
6 pm 7 pm
2 3
11/15/16
Vive SolRestaurant
MexicanCuisine
6:30 pm 7 pm
11/15/16Date
Time
Restaurants
7:30 pm
Century
16
Trolls
8 pm 9 pm
Movies

76
Dialog State Tracking Challenge (DSTC)
(Williams et al. 2013, Henderson et al. 2014, Henderson et al. 2014, Kim et al. 2016, Kim et al. 2016)
Challenge Type Domain Data Provider Main Theme
DSTC1 Human-Machine Bus Route CMU Evaluation Metrics
DSTC2 Human-Machine Restaurant U. Cambridge User Goal Changes
DSTC3 Human-Machine Tourist Information U. Cambridge Domain Adaptation
DSTC4 Human-Human Tourist Information I2R Human Conversation
DSTC5 Human-Human Tourist Information I2R Language Adaptation

77
NN-Based DST (Henderson et al., 2013; Henderson et al., 2014; Mrkšić et al., 2015;
Mrkšić et al., 2016)
77(Figure from Wen et al, 2016)
http://www.anthology.aclweb.org/W/W13/W13-4073.pdf; https://arxiv.org/abs/1506.07190; https://arxiv.org/abs/1606.03777

78
Neural Belief Tracker (Mrkšić et al., 2016)
78
https://arxiv.org/abs/1606.03777

79
Multichannel Tracker (Shi et al., 2016)
79
 Training a multichannel CNN for each slot
 Chinese character CNN
 Chinese word CNN
 English word CNN

80
DST Evaluation
 Dialogue State Tracking Challenges
 DSTC2-3, human-machine
 DSTC4-5, human-human
 Metric
 Tracked state accuracy with respect to user goal
 Recall/Precision/F-measure individual slots
80

81
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
81

82
Elements of Dialogue Management
82(Figure from Gašić)
Dialogue Policy Optimization

83
Dialogue Policy Optimization
 Dialogue management in a RL framework
83
U s e r
Reward R Observation OAction A
Environment
Agent
Natural Language Generation Language Understanding
Dialogue Manager
Slides credited by Pei-Hao Su
Optimized dialogue policy selects the best action that can maximize the future reward.
Correct rewards are a crucial factor in dialogue policy training

84
Reward for RL ≅ Evaluation for System
 Dialogue is a special RL task
 Human involves in interaction and rating (evaluation) of a dialogue
 Fully human-in-the-loop framework
 Rating: correctness, appropriateness, and adequacy
- Expert rating high quality, high cost
- User rating unreliable quality, medium cost
- Objective rating Check desired aspects, low cost
84

85
Reinforcement Learning for Dialogue Policy Optimization
85
Language
understanding
Language
(response)
generation
Dialogue
Policy
𝑎 = 𝜋(𝑠)
Collect rewards
(𝑠, 𝑎, 𝑟, 𝑠’)
Optimize
𝑄(𝑠, 𝑎)
User input (o)
Response
𝑠
𝑎
Type of Bots State Action Reward
Social ChatBots Chat history System Response
# of turns maximized;
Intrinsically motivated reward
InfoBots (interactive Q/A)
User current
question + Context
Answers to current
question
Relevance of answer;
# of turns minimized
Task-Completion Bots
User current input +
Context
System dialogue act w/
slot value (or API calls)
Task success rate;
# of turns minimized
Goal: develop a generic deep RL algorithm to learn dialogue policy for all bot categories

86
Dialogue Reinforcement Learning Signal
 Typical reward function
 -1 for per turn penalty
 Large reward at completion if successful
 Typically requires domain knowledge
✔ Simulated user
✔ Paid users (Amazon Mechanical Turk)
✖ Real users
|||
…
﹅
86
The user simulator is usually required for dialogue
system training before deployment

87
Neural Dialogue Manager (Li et al., 2017)
 Deep Q-network for training DM policy
 Input: current semantic frame observation, database returned results
 Output: system action
Semantic Frame
request_movie
request_location
DQN-based
Dialogue
Management
(DM)
Simulated User
Backend DB

88
SL + RL for Sample Efficiency (Su et al., 2017)
 Issue about RL for DM
 slow learning speed
 cold start
 Solutions
 Sample-efficient actor-critic
 Off-policy learning with experience replay
 Better gradient update
 Utilizing supervised data
 Pretrain the model with SL and then fine-tune with RL
 Mix SL and RL data during RL learning
 Combine both
88
https://arxiv.org/pdf/1707.00130.pdfSu et.al., SIGDIAL 2017

89
Online Training (Su et al., 2015; Su et al., 2016)
 Policy learning from real users
 Infer reward directly from dialogues (Su et al., 2015)
 User rating (Su et al., 2016)
 Reward modeling on user binary success rating
Reward
Model
Success/Fail
Embedding
Function
Dialogue
Representation
Reinforcement
SignalQuery rating
http://www.anthology.aclweb.org/W/W15/W15-46.pdf; https://www.aclweb.org/anthology/P/P16/P16-1230.pdf

90
Interactive RL for DM (Shah et al., 2016)
90
Immediate
Feedback
https://research.google.com/pubs/pub45734.html
Use a third agent for providing interactive feedback to the DM

91
Interpreting Interactive Feedback (Shah et al., 2016)
91
https://research.google.com/pubs/pub45734.html

92
Dialogue Management Evaluation
 Metrics
 Turn-level evaluation: system action accuracy
 Dialogue-level evaluation: task success rate, reward
92

93
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth
93

94
Natural Language Generation (NLG)
 Mapping semantic frame into natural language
inform(name=Seven_Days, foodtype=Chinese)
Seven Days is a nice Chinese restaurant
94

95
Template-Based NLG
 Define a set of rules to map frames to NL
95
Pros: simple, error-free, easy to control
Cons: time-consuming, poor scalability
Semantic Frame Natural Language
confirm() “Please tell me more about the product your are
looking for.”
confirm(area=$V) “Do you want somewhere in the $V?”
confirm(food=$V) “Do you want a $V restaurant?”
confirm(food=$V,area=$W) “Do you want a $V restaurant in the $W.”

96
Plan-Based NLG (Walker et al., 2002)
 Divide the problem into pipeline
 Statistical sentence plan generator (Stent et al., 2009)
 Statistical surface realizer (Dethlefs et al., 2013; Cuayáhuitl et al., 2014; …)
Inform(
name=Z_House,
price=cheap
)
Z House is a
cheap restaurant.
Pros: can model complex linguistic structures
Cons: heavily engineered, require domain knowledge
Sentence
Plan
Generator
Sentence
Plan
Reranker
Surface
Realizer
syntactic tree

97
Class-Based LM NLG (Oh and Rudnicky, 2000)
 Class-based language modeling
 NLG by decoding
97
Pros: easy to implement/ understand, simple rules
Cons: computationally inefficient
Classes:
inform_area
inform_address
…
request_area
request_postcode
http://dl.acm.org/citation.cfm?id=1117568

98
Phrase-Based NLG (Mairesse et al, 2010)
Semantic
DBN
Phrase
DBN
Charlie Chan is a Chinese Restaurant near Cineworld in the centre
d d
Inform(name=Charlie Chan, food=Chinese, type= restaurant, near=Cineworld, area=centre)
98
Pros: efficient, good performance
Cons: require semantic alignments
realization phrase semantic stack

99
RNN-Based LM NLG (Wen et al., 2015)
<BOS> SLOT_NAME serves SLOT_FOOD .
<BOS> Din Tai Fung serves Taiwanese .
delexicalisation
Inform(name=Din Tai Fung, food=Taiwanese)
0, 0, 1, 0, 0, …, 1, 0, 0, …, 1, 0, 0, 0, 0, 0…
dialogue act 1-hot
representation
SLOT_NAME serves SLOT_FOOD . <EOS>
Slot weight tying
conditioned on
the dialogue act
Input
Output
http://www.anthology.aclweb.org/W/W15/W15-46.pdf#page=295

100
Handling Semantic Repetition
 Issue: semantic repetition
 Din Tai Fung is a great Taiwanese restaurant that serves Taiwanese.
 Din Tai Fung is a child friendly restaurant, and also allows kids.
 Deficiency in either model or decoding (or both)
 Mitigation
 Post-processing rules (Oh & Rudnicky, 2000)
 Gating mechanism (Wen et al., 2015)
 Attention (Mei et al., 2016; Wen et al., 2015)
100

101
 Original LSTM cell
 Dialogue act (DA) cell
 Modify Ct
Semantic Conditioned LSTM (Wen et al., 2015)
DA cell
LSTM cell
Ct
it
ft
ot
rt
ht
dtdt-1
xt
xt ht-1
xt ht-1 xt ht-1 xt ht-
1
ht-1
Inform(name=Seven_Days, food=Chinese)
0, 0, 1, 0, 0, …, 1, 0, 0, …, 1, 0, 0, …
dialog act 1-hot
representation
d0
101
Idea: using gate mechanism to control the
generated semantics (dialogue act/slots)
http://www.aclweb.org/anthology/D/D15/D15-1199.pdf

102
Structural NLG (Dušek and Jurčíček, 2016)
 Goal: NLG based on the syntax tree
 Encode trees as sequences
 Seq2Seq model for generation
102
https://www.aclweb.org/anthology/P/P16/P16-2.pdf#page=79

103
Contextual NLG (Dušek and Jurčíček, 2016)
 Goal: adapting users’ way of
speaking, providing context-
aware responses
 Context encoder
 Seq2Seq model
103
https://www.aclweb.org/anthology/W/W16/W16-36.pdf#page=203

104
Controlled Text Generation (Hu et al., 2017)
 Idea: NLG based on generative adversarial network (GAN) framework
 c: targeted sentence attributes

105
NLG Evaluation
 Metrics
 Subjective: human judgement (Stent et al., 2005)
 Adequacy: correct meaning
 Fluency: linguistic fluency
 Readability: fluency in the dialogue context
 Variation: multiple realizations for the same concept
 Objective: automatic metrics
 Word overlap: BLEU (Papineni et al, 2002), METEOR, ROUGE
 Word embedding based: vector extrema, greedy matching,
embedding average
105
There is a gap between human perception and automatic metrics

107
Dialogue System Evaluation
107
 Dialogue model evaluation
 Crowd sourcing
 User simulator
 Response generator evaluation
 Word overlap metrics
 Embedding based metrics

108
Crowdsourcing for Dialogue System Evaluation (Yang et al., 2012)
108
http://www-scf.usc.edu/~zhaojuny/docs/SDSchapter_final.pdf
The normalized mean scores of Q2
and Q5 for approved ratings in each
category. A higher score maps to a
higher level of task success

109
User Simulation
 Goal: generate natural and reasonable conversations to enable reinforcement
learning for exploring the policy space
 Approach
 Rule-based crafted by experts (Li et al., 2016)
 Learning-based (Schatzmann et al., 2006; El Asri et al., 2016, Crook and Marin, 2017)
Dialogue
Corpus
Simulated User
Real User
• Dialogue Policy
Interaction
keeps a list of its goals
and actions
randomly generates an
agenda
updates its list of goals
and adds new ones

110
Elements of User Simulation
Error Model
• Recognition error
• LU error
Dialogue State
Tracking (DST)
System dialogue acts
Reward
Backend Action /
Knowledge Providers
Dialogue Policy
Optimization
User Model
Reward Model
User Simulation Distribution over
user dialogue acts
(semantic frames)
The error model enables the system to maintain the robustness

111
Rule-Based Simulator for RL Based System (Li et al., 2016)
111
 rule-based simulator + collected data
 starts with sets of goals, actions, KB, slot types
 publicly available simulation framework
 movie-booking domain: ticket booking and movie seeking
 provide procedures to add and test own agent

112
Model-Based User Simulators
 Bi-gram models (Levin et.al. 2000)
 Graph-based models (Scheffler and Young, 2000)
 Data Driven Simulator (Jung et.al., 2009)
 Neural Models (deep encoder-decoder)
112

113
Data-Driven Simulator (Jung et.al., 2009)
113
 Three step process
1) User intention simulator
Current
discourse
status (t-1)
User’s current
semantic
frame (t-1)
Current
discourse
status (t)
User’s current
semantic
frame (t)
Current
discourse
status
User’s current
semantic
frame
request+search_loc
(*) compute all possible semantic frame
given previous turn info
(*) randomly select one possible semantic frame
features (DD+DI)

114
114
 Three step process
2) User utterance simulator
request+search_loc
Given a list of POS tags associated with
the semantic frame, using LM+Rules
they generate the user utterance.
I want to go to the city hall
PRP VB TO VB TO [loc_name]

115
115
 Three step process:
2) User utterance simulator
3) ASR channel simulator
 Evaluate the generated sentences
using BLUE-like measures against
the reference utterances collected
from humans (with the same
goal)

116
Seq2Seq User Simulation (El Asri et al., 2016)
 Seq2Seq trained from dialogue data
 Input: ci encodes contextual features, such as the previous system action,
consistency between user goal and machine provided values
 Output: a dialogue act sequence form the user
 Extrinsic evaluation for policy

117
Seq2Seq User Simulation (Crook and Marin, 2017)
 Seq2Seq trained from dialogue data
 No labeled data
 Trained on just human to machine conversations

118
User Simulator for Dialogue Evaluation Measures
•whether constrained values specified by users can be understood by the system
•agreement percentage of system/user understandings over the entire dialog (averaging all turns)
Understanding Ability
•Number of dialogue turns
•Ratio between the dialogue turns (larger is better)
Efficiency
•an explicit confirmation for an uncertain user utterance is an appropriate system action
•providing information based on misunderstood user requirements
Action Appropriateness

119
How NOT to Evaluate Dialog System (Liu et al., 2017)
 How to evaluate the quality of the generated response ?
 Specifically investigated for chat-bots
 Crucial for task-oriented tasks as well
 Metrics:
 Word overlap metrics, e.g., BLEU, METEOR, ROUGE, etc.
 Embeddings based metrics, e.g., contextual/meaning
representation between target and candidate

120
Dialogue Response Evaluation (Lowe et al., 2017)
Towards an Automatic Turing Test
 Problems of existing automatic evaluation
 can be biased
 correlate poorly with human judgements of response
quality
 using word overlap may be misleading
 Solution
 collect a dataset of accurate human scores for variety
of dialogue responses (e.g., coherent/un-coherent,
relevant/irrelevant, etc.)
 use this dataset to train an automatic dialogue
evaluation model – learn to compare the reference to
candidate responses!
 Use RNN to predict scores by comparing against human
scores!
Context of Conversation
Speaker A: Hey, what do you want
to do tonight?
Speaker B: Why don’t we go see a
movie?
Model Response
Nah, let’s do something active.
Reference Response
Yeah, the film about Turing looks
great!

End-to-End Learning for Dialogues
Multimodality
Dialogue Breath
Dialogue Depth
Recent Trends and Challenges121

122
Outline
 Introduction
 Evaluation
 Multimodality
 Dialogue Breath
 Dialogue Depth

123
ChitChat Hierarchical Seq2Seq (Serban et al., 2016)
 Learns to generate dialogues from offline dialogs
 No state, action, intent, slot, etc.
http://www.aaai.org/ocs/index.php/AAAI/AAAI16/paper/view/11957

124
ChitChat Hierarchical Seq2Seq (Serban et.al., 2017)
 A hierarchical seq2seq model with Gaussian latent variable for generating
dialogues (like topic or sentiment)

125
Knowledge Grounded Neural Conv. Model (Ghazvininejad et al.,
2017)
125

126
E2E Joint NLU and DM (Yang et al., 2017)
 Errors from DM can be propagated to NLU for
regularization + robustness
DM
126
Model DM NLU
Baseline (CRF+SVMs) 7.7 33.1
Pipeline-BLSTM 12.0 36.4
JointModel 22.8 37.4
Both DM and NLU performance (frame accuracy) is improved

127
0 0 0 … 0 1
Database Operator
Copy
field
…
Database
Sevendays
CurryPrince
Nirala
RoyalStandard
LittleSeuol
DB pointer
Can I have korean
Korean
0.7
British
0.2
French
0.1
…
Belief Tracker
Intent Network
Can I have <v.food>
E2E Supervised Dialogue System (Wen et al., 2016)
Generation Network
<v.name> serves great <v.food> .
Policy Network
127
zt
pt
xt
MySQL query:
“Select * where
food=Korean”
qt

128
E2E MemNN for Dialogues (Bordes et al., 2016)
 Split dialogue system actions into
subtasks
 API issuing
 API updating
 Option displaying
 Information informing

129
E2E RL-Based KB-InfoBot (Dhingra et al., 2017)
Movie=?; Actor=Bill Murray; Release Year=1993
Find me the Bill Murray’s movie.
I think it came out in 1993.
When was it released?
Groundhog Day is a Bill Murray
movie which came out in 1993.
KB-InfoBot
User
Entity-Centric Knowledge Base
129Idea: differentiable database for propagating the gradients
http://www.aclweb.org/anthology/P/P17/P17-1045.pdf
Movie Actor
Release
Year
Groundhog Day Bill Murray 1993
Australia Nicole Kidman X
Mad Max: Fury Road X 2015

130
E2E RL-Based System (Zhao and Eskenazi, 2016)
130
 Joint learning
 NLU, DST, Dialogue Policy
 Deep RL for training
 Deep Q-network
 Deep recurrent network
Baseline
RL
Hybrid-RL
http://www.aclweb.org/anthology/W/W16/W16-36.pdf

131
E2E LSTM-Based Dialogue Control (Williams and Zweig, 2016)
131
 Idea: an LSTM maps from raw dialogue history directly to a distribution
over system actions
 Developers can provide software including business rules & programmatic APIs
 LSTM can take actions in the real world on behalf of the user
 The LSTM can be optimized using SL or RL

132
E2E Task-Completion Bot (TC-Bot) (Li et al., 2017)
wi
B-
type
wi
+1
wi+2
O O
EOS
<intent
>
wi
B-
type
wi
+1
wi+2
O O
EOS
<intent
>
Semantic Frame
request_movie
genre=action,
date=this weekend
System Action /
Policy
request_location
User Dialogue Action
Inform(location=San Francisco)
Time t-1
wi
<slot>
wi
+1
wi+2
O O
EOS
<intent>
Time t-2
Time t
Dialogue
Management
(DM)
w0 w1 w2
Natural Language Generation (NLG)
EOS
User
Goal
User Agenda Modeling
User Simulator
End-to-End Neural Dialogue System
Text Input
Are there any action movies
to see this weekend?
Idea: supervised learning for each component and reinforcement learning for
end-to-end training the neural dialogue system
132

133
E2E Task-Completion Bot (TC-Bot) (Li et al., 2017)
 User goal: Two tickets for “the witch” tomorrow 9:30 PM at regal meridian 16, Seattle.
RULE BASED SYSTEM
User: Can I get 2 tickets for the witch?
Agent: Which theater do you want?
User: Regal meridian 16
Agent: What time would you like to see it?
User: 9:30 pm
Agent: What date would you like to see it?
User: Tomorrow!
Agent: How many people are you?
User: Just two adults.
Agent: What is your location?
User: Seattle!
Agent: What kind of video format do you like?
User: Oh, any kind is okay.
Agent: Do you have any price constraint?
User: No, I don’t care.
Agent: Great - I was able to purchase 2 tickets for
you to see the witch tomorrow at regal meridian 16
theater in Seattle at 9:30 pm.
User: Thank you.
Agent: Thank you! 133
The system can learn how to efficiently
interact with users for task completion
REINFORCEMENT LEARNING SYSTEM
User: Can I get 2 tickets for the witch?
Agent: What date would you like to see it?
User: Tomorrow
Agent: Which theater do you prefer?
User: Regal meridian 16
Agent: What is your location?
User: Seattle
Agent: What time would you like to see it?
User: 9:30 pm.
Agent: Great - I was able to purchase 2 tickets for
you to see the witch tomorrow at regal meridian
16 theater in Seattle at 9:30 pm.
User: Thanks.
Agent: Thanks!

134
Hierarchical RL for Composite Tasks (Peng et al., 2017)
134
Travel Planning
Actions
• Set of tasks that need to be fulfilled collectively!
• Build a dialog manager that satisfies cross-
subtask constraints (slot constraints)
• Temporally constructed goals
• hotel_check_in_time > departure_flight_time
• # flight_tickets = #people checking in the hotel
• hotel_check_out_time< return_flight_time,
https://arxiv.org/abs/1704.03084Peng et.al., EMNLP 2017

135
Hierarchical RL for Composite Tasks (Peng et al., 2017)
135
 The dialog model makes decisions over two levels: meta-
controller and controller
 The agent learns these policies simultaneously
 the policy of optimal sequence of goals to follow 𝜋 𝑔 𝑔𝑡, 𝑠𝑡; 𝜃1
 Policy 𝜋 𝑎,𝑔 𝑎 𝑡, 𝑔𝑡, 𝑠𝑡; 𝜃2 for each sub-goal 𝑔𝑡
Meta-
Controller
Controller
(mitigate reward sparsity issues)
https://arxiv.org/abs/1704.03084Peng et.al., EMNLP 2017

136
Outline
 Introduction
 Multimodality
 Dialogue Breath
 Dialogue Depth
136

137
Brain Signal for Understanding
137
 Misunderstanding detection by brain signal
 Green: listen to the correct answer
 Red: listen to the wrong answer
Detecting misunderstanding via brain signal in order to correct the understanding results

138
Video for Intent Understanding
138
Proactive (from camera)
I want to see a movie on TV!
Intent: turn_on_tv
May I turn on the TV for you?
Proactively understanding user intent to initiate the dialogues.

139
App Behavior for Understanding
139
 Task: user intent prediction
 Challenge: language ambiguity
 User preference
✓ Some people prefer “Message” to “Email”
✓ Some people prefer “Ping” to “Text”
 App-level contexts
✓ “Message” is more likely to follow “Camera”
✓ “Email” is more likely to follow “Excel”
send to vivian
v.s.
Email? Message?
Communication
Considering behavioral patterns in history to model understanding for intent prediction.

140
Video Highlight Prediction Using Audience Chat Reactions
140
https://arxiv.org/pdf/1707.08559.pdfFu et.al., EMNLP 2017

141
Video Highlight Prediction Using Audience Chat Reactions
141
https://arxiv.org/pdf/1707.08559.pdfFu et.al., EMNLP 2017
 Goal: predict highlight from the video
 Input : multi-modal and multi-lingual
(real time text commentary from fans)
 Output: tag if a frame part of a highlight
or not

142
Evolution Roadmap
142
Dialogue breadth (coverage)
Dialoguedepth(complexity)
What is influenza?
I’ve got a cold what do I do?
Tell me a joke.
I feel sad…

143
Outline
 Introduction
 Multimodality
 Dialogue Breath
 Dialogue Depth
143

144
Evolution Roadmap
144
Single
domain
systems
Extended
systems
Multi-
domain
systems
Open
domain
systems
What is influenza?
Tell me a joke.
I feel sad…

145
Intent Expansion (Chen et al., 2016)
 Transfer dialogue acts across domains
 Dialogue acts are similar for multiple domains
 Learning new intents by information from other domains
CDSSM
New Intent
Intent Representation
1
2
K
:
Embedding
Generation
K+1
K+2<change_calender>
Training Data
<change_note>
“adjust my note”
:
<change_setting>
“volume turn down”
The dialogue act representations can be automatically learned for other domains
postpone my meeting to five pm

146
Zero-Shot Learning (Daupin et al., 2016)
 Semantic utterance classification
 Use query click logs to define a task that makes the networks learn the meaning or
intent behind the queries
 The semantic features are the last hidden layer of the DNN
 Use Zero-Shot Discriminative embedding model combines H with the minimization
of entropy of a zero-shot classifier

147
Domain Adaptation for SLU (Kim et al., 2016)
 Frustratingly easy domain adaptation
 Novel neural approaches to domain adaptation
 Improve slot tagging on several domains
http://www.aclweb.org/anthology/C/C16/C16-1038.pdf

148
Policy for Domain Adaptation (Gašić et al., 2015)
 Bayesian committee machine (BCM) enables estimated Q-function to
share knowledge across domains
QRDR
QHDH
QL DL
Committee Model
The policy from a new domain can be boosted by the committee policy

149
Outline
 Introduction
 Multimodality
 Dialogue Breath
 Dialogue Depth
149

150
Evolution Roadmap
150
Knowledge based system
Common sense system
Empathetic systems
What is influenza?
Tell me a joke.
I feel sad…

151
High-Level Intention for Dialogue Planning (Sun et al., 2016)
 High-level intention may span several domains
Schedule a lunch with Vivian.
find restaurant check location contact play music
What kind of restaurants do you prefer?
The distance is …
Should I send the restaurant information to Vivian?
Users can interact via high-level descriptions and the system learns how to plan the dialogues
http://dl.acm.org/citation.cfm?id=2856818; http://www.lrec-conf.org/proceedings/lrec2016/pdf/75_Paper.pdf

152
Empathy in Dialogue System (Fung et al., 2016)
 Embed an empathy module
 Recognize emotion using multimodality
 Generate emotion-aware responses
152Emotion Recognizer
vision
speech
text

153
Visual Object Discovery through Dialogues (Vries et al., 2017)
 Recognize objects using “Guess What?” game
 Includes “spatial”, “visual”, “object taxonomy” and “interaction”
153

155
Summarized Challenges
155
Human-machine interfaces is a hot topic but several components must be integrated!
Most state-of-the-art technologies are based on DNN
• Requires huge amounts of labeled data
• Several frameworks/models are available
Fast domain adaptation with scarse data + re-use of rules/knowledge
Handling reasoning
Data collection and analysis from un-structured data
Complex-cascade systems requires high accuracy for working good as a whole

156
Brief Conclusions
 Introduce recent deep learning methods used in dialogue models
 Highlight main components of dialogue systems and new deep
learning architectures used for these components
 Talk about challenges and new avenues for current state-of-the-art
research
 Provide all materials online!
156
http://deepdialogue.miulab.tw

THANKS FOR ATTENTION!
Q & A
Thanks to Tsung-Hsien Wen, Pei-Hao Su, Li Deng, Jianfeng Gao,
Sungjin Lee, Milica Gašić, Lihong Li, Xiujin Li, Abhinav Rastogi, Ankur
Bapna, PArarth Shah and Gokhan Tur for sharing their slides.
deepdialogue.miulab.tw

2017 Tutorial - Deep Learning for Dialogue Systems

More Related Content

What's hot (20)

Similar to 2017 Tutorial - Deep Learning for Dialogue Systems (20)

More from MLReview (13)

Recently uploaded (20)

2017 Tutorial - Deep Learning for Dialogue Systems