Thursday, November 29, 2018

We are second again


We have made it again with the Alquist bot. Alquist competed with the Alana bot from the Heriot-Watt University, Edinburgh, Scotland and the winning Gunrock bot from the University of California Davis Davis, CA, USA.

A $500,000 prize was awarded to the winning team. We are bringing back $100,000 in prize money, and Alana receives $50,000, The challenge for additional, a $1 million research grant has not been awarded yet. It will take some time to make 20 minutes long chat. Just imagine how difficult it will be to get to a bar and talk to a stranger for 20 min.

The Alexa Prize is a $3.5 million challenge for university teams to advance human-computer interaction. Similarly, as last year the goal was to develop the best social bot conversing coherently and engagingly with humans on a range of current events and favorite topics such as entertainment, sports, politics, technology, and fashion. We continued this year with the Alquist II starting from the beginning of 2018 when the Amazon Alexa Prize was announced. We submitted our proposal, and we have made it in between eight semifinalists who were selected from more than a hundred teams from 15 countries.  Amazon has awarded us with a $250,000 research grant, Alexa-enabled devices, and free Amazon Web Services (AWS) to support our development efforts.

The research grant was significant support for our team The team leader as last year was Jan Pichl who is pursuing the third year of his Ph.D. program at the faculty of Electrical Engineering in Conversational AI. This year charged with enthusiasm the team decided to drop the first version of Alquist and started from scratch with an entirely superior, new design. We built on the latest neural network technology in combination with a small number of rules to conduct the dialogs. Alquist II knows how to react to most conversational utterances, but it excels in 26 selected topics. A great deal in the quality improvement came from a large number of users conversing with our bot. Each conversation helps to understand better the complexity and select the best matching reply.

In-depth knowledge is required to create an exciting and entertaining conversation. Where to get the content? The web is an endless source of interesting facts, but mostly in a written text. When played back it feels too long and a little tedious. To make the conversation natural, we had to solve this problem. The large part of work went to the knowledge acquisition and processing. If you are a lucky owner of Alexa device, you can test the Alquist abilities, just say let’s chat with Alquist.

The Alexa Price winners announcement was part of the AWS re:invent conference in Las Vegas. The finalists were invited. We all have enjoyed a grand celebration as well as the gathering.

Amazon is investing a lot in the development of intelligent conversational gadgets led by Alexa. The experts predict that the most natural way for communication, the speech will become in the nearest future an additional channel to control appliances, access knowledge, etc. It is fascinating and inspiring to find our team between the leading groups in the world working on the latest technology with an exciting vision. We wish our success will attract new students to join our team an pursue our adventure next year. Let us know!


Friday, August 31, 2018

We are the Alexa Prize finalists again


We have made it to the Alexa Prize 2018 finals again with our social bot Alquist. Our competitors are the Alana bot from the Heriot-Watt University, Edinburgh, Scotland and the Gunrock bot from the UC Davis, Davis, CA.

It was almost exactly one year ago I wrote the last blog. That time we were excited to get to the Alexa Prize 2017 finals, and we celebrate today again, we made it to the finals with the Alquist team again. It was a hectic time.

We have completely redesigned our bot. This year when we started the semifinals, we experienced problems with data to train the new AI. As the number of interactions was growing, we were increasing the training sets and improving the accuracy. We have augmented the dialog acts classifier processing every new user utterance. It is using the convolutional neural network and classifies the utterances to around thirty classes. The significant change in the overall architecture is the dialog manager. Last year we used a rule-based approach. It was great for cooperative users, but once the user did something unexpected we had troubles. It was also a very laborious process to write the rules. We ended with hundreds and hundreds of rules. It was also challenging to update or enhance the dialogs.  The latest Alquist uses hybrid dialog management. We have reduced the rule-based decision to a minimum and made the principal part controlled by an LSTM neural network. We have many LSTM models for different sub-dialogs. The sub-dialogs are trained and updated for excerpts of the bot user interactions. The hybrid approach significantly reduced the amount of work necessary to create a new dialogue compared to last year's rule-based approach. This fact allowed us to broaden the range of conversational topics substantially. We have also taken advantage of delexicalizing the utterances to improve the training speed. The bot includes several other neural networks helping to switch between different topics, estimating the sentiment, etc. The whole system is getting quite complicated. We have also spend a lot of efforts on improving the new information acquisition. We are crawling several social media. The discussions are an additional source of interesting facts. The social media are a great complement to knowledge databases with the factoid type of information like for example Wikipedia.

The team has changed a little compared to the last year. Roman has left, and Petr Lorenc has joined. He is helping a lot with the intent, entity recognition, which is an essential part of Alquist and has a significant impact on the overall user experience. Currently, everybody is very busy since we have another at least two months to improve the functionality. We will focus on the user experience. Since English is not our native language, we have to spend a lot of effort ironing out all conversation, adding SSML, etc.  Amazon will offer to the Alexa device owners only three first bots, which means we will get more data. More data gives us a chance to improve further the accuracy.

Amazon will announce the winners as last year at the re:Invent Amazon Conference in Las Vegas. We are looking forward to visiting Las Vegas the heart of gambling, as well as meeting our competitors and helpful Amazon Alexa Prize staff, as well as learning the latest from the Amazon technology. We were second behind the Washington team last year. Guess what are our aspirations this year. If you are a lucky Alexa device owner, try "Alexa let's chat." Keep the fingers crossed!