Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… ·...
Transcript of Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… ·...
![Page 1: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/1.jpg)
Generative Deep Neural Networks for Dialogue
Presented By Shantanu Kumar
Adapted from slides by Iulian Vlad Serban
![Page 2: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/2.jpg)
What are Dialogue Systems?• Computer system that can
converse like a human with another human while making sense
• Types of Dialogue • Open Domain • Task Oriented
![Page 3: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/3.jpg)
Applications of Dialogue Systems• Technical Support
• Product enquiry • Website navigation • HR helpdesk • Error diagnosis • IVR system in Call Centres
• Entertainment • IoT interface • Virtual Assistants
• Siri, Cortana, Google Assistant • Assistive technology • Simulate human conversations
![Page 4: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/4.jpg)
How do we build such a system??
![Page 5: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/5.jpg)
Traditional Pipeline models
![Page 6: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/6.jpg)
End-To-End models with DL
Neural Network ResponseDialogue Context
![Page 7: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/7.jpg)
End-To-End models with DL
Knowledge Database
Actions
![Page 8: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/8.jpg)
What is a good Chatbot?The responses should be
• Grammatical • Coherent • In Context
• Ideally non-Generic responses
![Page 9: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/9.jpg)
How can we learn the model?
- Unsupervised Learning (Generative Models) - Maximise likelihood w.r.t. words
- Supervised Learning - Maximise likelihood w.r.t. annotated labels
- Reinforcement Learning - Learning from real users - Learning from simulated users - Learning with given reward function
![Page 10: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/10.jpg)
Generative Dialogue ModelingDecomposing Dialogue Probability,
Decomposing Utterance Probability,
![Page 11: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/11.jpg)
Maximising likelihood on fixed corpora - Imitating human dialogues
Generative Dialogue Modeling
![Page 12: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/12.jpg)
Models proposed with three inductive biases
• Long-term memory - Recurrent units used (GRU)
• High-level compositional structure - Hierarchical structure - Multi resolution representation (MRRNN paper)
• Representing uncertainty and ambiguity - Latent variables (MRRNN and VHRED)
Generative Dialogue Modeling
![Page 13: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/13.jpg)
Hierarchical Recurrent Encoder-Decoder (HRED)
- Encoder RNN - For encoding each utterance independently into an utterance vector
- Context RNN - For encoding the topic/context of the dialogue up till the current utterance using utterance vectors
- Decoder RNN - For predicting the next utterance
Akshay: Can be applied to arbitrary lengths
![Page 14: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/14.jpg)
Hierarchical Recurrent Encoder-Decoder (HRED)
![Page 15: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/15.jpg)
Bidirectional HRED
- Encoder RNN -> Bidirectional
- Forward and Backward RNNs combined to get fixed length representation - Concat last state of each RNN - Concat of L2 pooling over temporal dimension
Hierarchical Recurrent Encoder-Decoder (HRED)
![Page 16: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/16.jpg)
Hierarchical Recurrent Encoder-Decoder (HRED)
Bootstrapping
- Initialising with Word2Vec embeddings - Trained on Google News dataset
- Pre-training on SubTle Q-A dataset - 5.5M Q-A pairs - Converted to 2-turn dialogue D = {U1 = Q, U2 = A}
Barun Akshay Prachi Dinesh Gagan
Prachi: 2 stage training
![Page 17: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/17.jpg)
Dataset - MovieTriples dataset
• Open Domain - Wide variety of topics covered
• Names and Numbers replaced with <person> and <number> tokens
• Vocab of 10K most popular tokens
• Special <continued-utterance> and <end-of-utterance> tokens to capture breaks
Gagan, Rishabh, Dinesh
Why only triples?
Anshul:
Split train/val on movies?
![Page 18: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/18.jpg)
Dialogue ModelingUbuntu Dialog Corpus - Goal-driven: Users resolve technical problems
- ~0.5M dialogues
Twitter Dialog Corpus - Open-domain: Social chit-chat
- ~0.75M dialogues in Train, 100K for Val and Test - 6.27 utterance and 94 tokens per dialogue
![Page 19: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/19.jpg)
Expert
Hello! Recently I updated to ubuntu 12.04 LTS and I am unsatisfied by its performance. I am facing a bug since the upgrade to 12.04 LTS. Can anyone help??????????
User
Example - Ubuntu Corpus
![Page 20: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/20.jpg)
Expert
Hello! Recently I updated to ubuntu 12.04 LTS and I am unsatisfied by its performance. I am facing a bug since the upgrade to 12.04 LTS. Can anyone help??????????
You need to give more details on the issue.
User
Example - Ubuntu Corpus
![Page 21: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/21.jpg)
Expert
Hello! Recently I updated to ubuntu 12.04 LTS and I am unsatisfied by its performance. I am facing a bug since the upgrade to 12.04 LTS. Can anyone help??????????
You need to give more details on the issue.
Every time I login it gives me "System Error" pop up. It is happing since I upgraded to 12.04.
User
Example - Ubuntu Corpus
![Page 22: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/22.jpg)
Expert
Hello! Recently I updated to ubuntu 12.04 LTS and I am unsatisfied by its performance. I am facing a bug since the upgrade to 12.04 LTS. Can anyone help??????????
You need to give more details on the issue.
Every time I login it gives me "System Error" pop up. It is happing since I upgraded to 12.04.
Send a report, or cancel it.
User
Example - Ubuntu Corpus
![Page 23: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/23.jpg)
Example - Ubuntu CorpusExpert
Hello! Recently I updated to ubuntu 12.04 LTS and I am unsatisfied by its performance. I am facing a bug since the upgrade to 12.04 LTS. Can anyone help??????????
You need to give more details on the issue.
Every time I login it gives me "System Error" pop up. It is happing since I upgraded to 12.04.
Send a report, or cancel it.
I have already done that but after few min, it pops up again...
User
![Page 24: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/24.jpg)
Example - Twitter CorpusPerson B
Hanging out in the library for the past couple hours makes me feel like I'll do great on this test! @smilegirl400 wow, what a nerd lol jk haha
=p what!? you changed your bio =(
@smileman400 Do you like my bio now? I feel bad for changing it but I like change. =P @smilegirl400 yes I do =) It definitely sums
up who you are lisa. Yay! you still got me =)
Person A
![Page 25: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/25.jpg)
Evaluation Metric- Word Perplexity - Measures the probability of generating the exact reference utterance
- Word error-rate - Number of words in the dataset the model has predicted incorrectly divided by the total number of words in the dataset. - Penalises diversity [Akshay]
Barun Akshay Dinesh Rishabh Arindam Anshul
![Page 26: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/26.jpg)
- Word Perplexity - Can only be used with generative models - Given an utterance, what is the probability?
How do we evaluate given an output utterance? - Multi-modal output - Space of possible valid utterance is huge - Human annotation is expensive and slow
Evaluation Metric
![Page 27: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/27.jpg)
How do we evaluate given an output utterance? - Multi-modal output - Space of possible valid utterance is huge - Human annotation is expensive and slow
Automatic Evaluation Metrics - Word overlap measure (BLEU, ROUGE, Levenshtein dist.) - Embedding based measures - Poor correlation with Human annotation
Evaluation Metric
![Page 28: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/28.jpg)
Results
Lack of error analysis
![Page 29: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/29.jpg)
MAP Output- Most probable last utterance - Found using beam search for better approximation - Generic responses observed - Stochastic sampling gives more diverse dialogues
Nupur: MAP vs Stochastic Sampling
![Page 30: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/30.jpg)
![Page 31: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/31.jpg)
ExtensionsModel • [Barun][Rishabh] Attention model during decoding for long
contexts • [Prachi] Dialogue systems with multiple participants
• Different decoders for each participant? • Order of speaking
• [Rishabh] Incorporating outside knowledge using KB
![Page 32: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/32.jpg)
ExtensionsData • [Akshay][Surag] Use bigger datasets like Reddit for dialogue • [Rishabh] Using film dialogue scripts from films like "Ek ruka
hua fasla" might be useful. • [Barun] Artificially scoring generic responses • [Surag] Prune generic responses from training data
![Page 33: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/33.jpg)
Extensions• [Prachi] Automatic generation of dialogue for movie given
storyline and character description • [Gagan] Pre-train word embeddings on SubTle • [Arindam] RL is the best bet to avoid generic responses • [Arindam] Adversarial evaluation • [Arindam] Train additional context to add consistency?
![Page 34: Generative Deep Neural Networks for Dialoguemausam/courses/col864/spring2017/slides/15-h… · Generative Deep Neural Networks for Dialogue Presented By Shantanu Kumar Adapted from](https://reader034.fdocuments.us/reader034/viewer/2022042220/5ec6c60f2e26f1010c6d8d67/html5/thumbnails/34.jpg)
Thank You