Archives de
Author: Bonenfant Guillaume

My daily life in Japan (as a student)

My daily life in Japan (as a student)

My daily life as a student in Japan and as a student in France (and even in Sweden) were very different and so I’ve decided to give a little part of my daily life there.
But before getting into the main part, I wanna let know that this might not be a « normal » life as a student in Japan. We, exchange students, had a special treatment in a lot of ways that anyone doesn’t have, like a cafeteria in our dormitory and a great time table.

Firstly, about my student apartment, I had a very small room in a dormitory that simply allowed me to sleep, work and store a very few things and even I, who doesn’t need a lot of space, felt a little bit cramped in there but it was enough for a 3-month time in Japan. Moreover, I had a balcony if I wanted to relax a little with fresh air.

I could start everyday (except when it was closed on sundays and holidays) with a great breakfast already prepared for students living in this dormitory. I personally liked a lot food in the dormitory, first we didn’t have to prepare it, second it was very healthy and third I could discover a lot of food that was unknown to me (and is still unknown to me 😂).

Then, I had about a 40 minutes ride to the university which started at 9AM.
University courses ended mostly around 2:30PM and then we had a lot of free time which I’m thankful for because we have been able to spend some great time, students and volunteering students, together and do a lot of activities thanks to it.

Our « typical » following after the courses was going to a cafe called « St.Marc Cafe » and eat delicious French food with for example the unfamous Choco Croissant.

Joke aside, St.Marc Cafe’s food is very good and I recommend it even though it’s not real French food. It doesn’t need to be French to be good !

When not going to a cafe after the courses, we could also sometimes just go somewhere else to visit or do activities, like we’ve once done baseball in a batting center for example. Elseway, we could just head back home to do some homeworks and then do things on our own, whatever things we like !

Again, in the evening, we had the luck to get our dinner prepared by the cafeteria which allowed for more time for having fun or working instead of cooking our food ourself !

For the things I liked to do personally, because we came in Japan in winter, I liked to go to places where there were illumations, beautiful light events in this period. When spring came, I instead went to some parks enjoying cherry blossoms. I got surprised how early they bloomed in the year, which I heard was pretty unusual, starting already from the end of february for a few trees in Tokyo.

Appart from these, I also went a lot to karaoke (my favorite activity in Japan !) and I’ve had the chance to be able to go to Fuji-Q Highland, a very well known amusement park with a lot of « world’s best attraction » !

To conclude, I loved every aspects of my student life in Japan, from the great Japanese courses we had in the university to also everything aside from it, the life and food in the dormitory, activities done with volounteers, by ourself exchange students or even the time spent alone where I could discover more Tokyo and the things I had the most interest in.

My Japanese learning experience in Nihon University (日本大学)

My Japanese learning experience in Nihon University (日本大学)

After my exchange semester in Sweden, I’ve directly taken the plane for another exchange semester, but this time way further to… Japan ! So from now on, I’m going to speak about this exchange semester, my experience in Tokyo and a few surroundings !

Let’s start today with what was the main goal of my semester in Japan : learning Japanese !

Main courses

During the Japanese learning, there were two types of courses, what I would call « main courses » which are Japanese language focused, and then cultural courses with a very wide range of subjects, about Japan and its culture.

Main courses were every morning from 9:00 to 12:10. These courses were very complete and thanks to them I could learn a lot on every aspects of the language, from grammar to kanji drawing, listening, writing…

Before coming to Japan, I learnt a little Japanese by myself and because I was always listening to Japanese songs and watching a lot of Japanese content, I knew already a lot of vocabulary and could make basic sentences and have conversations during everyday life. I managed for example to have a quite long conversation with a lady in the airplane going to Japan. It was a very funny moment, I have great memories of it, me trying for the first time to have a good conversation with a Japanese person, in Japanese.

Grammar being the aspect of the language I had the more struggles with in the beginning, I quickly realized how fast I made up for the gap with these courses and I could quickly made more complex sentences. I was also very bad at writing because I never had a good opportunity to practice in France, but again, with a good effort, it was possible to quickly learn the most important kanji that I should have already known according to my level.

Cultural courses

These courses which were focused on way broader things about Japan and its culture were every afternoon, from Monday to Thursday.

On Mondays, we had courses with a new teacher every 2 weeks, focusing on different subjects, economy, musics, the main issues of the country and how Japan could try to solve them, how do Japanese people communicate and the differences with our countries. These courses were all very interesting, with my favorite being the one about Japanese musics and its evolution through years (and also centuries), probably because it’s the Japan-related thing I like to talk about the most.

On Tuesdays we had older-culture related courses, going from shintoism to Japanese gardens and tea ceremony. We learnt a lot on people’s thoughts on shintoism thanks to volounteers, what are the main practices, where do they pray, what are the main elements in these sanctuaries, and if they see themselves as shintoists or not. About Japanese gardens, we learnt what are the different types of gardens (rock, tea, pond) and what we can find in them. We also studied tea ceremony, what people do there and how and the whole ritual around it and how meticulous people are during them. We also had the chance to participate in a few tea ceremonies after these courses.

On Wednesdays, we learnt about very broad topics among Japanese society and culture. We for example learnt about Japanese natural disasters like earthquakes, tsunami, Typhoons. We had the chance to go to a disaster prevention building(防災館)and to learn how to react in case any disaster would occur. Fun fact, I felt the first earthquake in my life on the day after this visit. During this course, we also talked about Japanese mindset and its origins, why they think more about the group than themselves. It was a very interesting course where I could learn a lot of things.

On Thursdays, we had autonomous learning courses. It’s a course where we could learn Japanese the way we wanted to with help from Japanese students. We were told that we could for example learn through musics, poems, movies, books and a lot more. I personally learnt most of the time using youtube videos from the « Nakata University » channel. Nakata is a youtuber talking about a very wide range of topics from self improvement tips to recent technologies or society. I felt that learning through his videos was very helpful as I could understand a good part of it and that I could learn and understand the other part thanks to the volounteers’ help.

School trips

Apart from these courses, we also had a few school trips in different places, Tokyo and aside. We had for example a trip to watch a Kabuki play in the evening which was very interesting to see even if it was hard understanding what was being said because of the old Japanese used in this kind of plays. Outside the Kabuki building, we took my favorite picture of my whole 3-month time in Japan. We also had a trip we prepared ouselves in Harajuku to be able to witness the old and modern sides of this district which is something that we can find a lot in Tokyo, as well as another one in Yokohama.

My favorite picture from my time in Japan

Conclusion

To conclude, I definitely could fulfill the main goal of this trip which was Japanese language learning as well as witnessing myself the cultural differences and to see if I liked the life there or not, and the answer is an undoubtful yes. I want to give a special thanks to everyone who was aside us from the university, from office workers who helped us a lot in everything we wanted to do, to teachers that were always very energetic, smiling and giving their all to help us as well as to volounteering Japanese students thanks to whom we could learn even more Japanese, build great friendships and spend some great time all together in Japan !

Weather and landscapes in Växjö

Weather and landscapes in Växjö

What I enjoyed the most during my semester in Växjö was nature and its beautiful landscapes. Växjö is a little city of about 71k inhabitants surrounded by several lakes and is known for being the greenest city of Europe.

Being in Sweden from the end of August until early January, I’ve been able to go through three different seasons, seeing the end of summer, fall and the early winter.

Summer

Beautiful landscape at a lake during sunset

Summer in Växjö feels pretty good with temperatures around 20 to 25°C. It’s a temperature where you can enjoy everything, from barbecues to sports or bathing in lakes. It was a good time for me to arrive in Sweden knowing what I would have to face in winter.

Picture taken on the very last day of summer.

Fall

Fall, like every other season, is a very beautiful moment in Växjö. Leafs start turning yellow, orange, red, and make it even more enjoyable to have a little run around the lakes.

Something special about Växjo and Sweden in general is that, at least from my experience, it is very often (not to say all the time) cloudy, and you also have a lot of rain that comes with it. Happily, the rain there is more often than not very thin and enjoyable.

If you’re going for a walk at night, you’ll be able to witness totally different landscapes in darkness, as much in the city as in more natural places.

Winter

Winter is the toughest season to live in Sweden but is also the one that fits this country best. Under snow, this city suddenly looks totally different. During my stay, temperatures went as low as -15°C, and it’s on this day that I could take among the best pictures. This trip in Sweden taught me that it was not unbearable for me to live in a that cold place, as long as I wear clothes warm enough. The wind being not too strong there, -15°C actually feels warmer (or let’s say, less cold) than I expected !

Once again, going for a walk by night can give you a totally different view of the city, this time under beautiful snow and lights.

Courses in Linnaeus University : Information Visualization

Courses in Linnaeus University : Information Visualization

During my exchange semester in Linnaeus University, I’ve had 5 different courses. I’ve particularly liked one of them, so I’ve decided to make a blog post about it, my most favorite course : Information Visualization.

In this course, we had the very basic goal of learning a lot about the different ways to visualize different kinds of information and being able in the end to think about our own representations.

Information visualization can take a very wide range of possibilities and there can be tons of visualizations for a single group of data. The example below shows perfectly the complexity an information display can take.

Fichier:Minard.png — Wikipédia
Charles Minard’s map of Napoleon’s disastrous Russian campaign of 1812

In the above’s map, Charles Minard represented the disastrous Russian invade campaign using six different types of data : the number of Napoleon’s troops, distance, temperature, the latitude and longitude, direction of travel, and location relative to specific dates. He indeed represented all these information in a very original way that might look at first a little bit hard to understand, but once we start to understand a little the map, we can get really quickly the picture out of it.

During this course, we have seen a lot of different visualization techniques like preattentive processing, data transformation using dynamic queries with for example range sliders, view transformations using user interaction with for example zoom and pan, just to cite a few.

At the end of the course, we had to make our own visualization of a specific set of data about cars and their characteristics. Mine was the following :

My visualization for the final project

As we had a lot of different characteristics for each car (26 features in total), I decided to display them using three different types of chart. At first a point chart, then a bar plot and finally a star plot.

The point chart is used to display and compare every cars using the numeric data. It is possible to change the features we are comparing them with using the simple select tools.

The bar plot is used to compare the types of cars and putting them in groups using text data in a way we can easily find what type of car is a specific point on the first chart with the use of colors.

The star plot is used to display every information of a specific car after clicking on it on the point chart. So for example if we feel interested in a car because it looks cheap with a high horsepower, we can also see easily its other statistics in the star plot so we can make up our mind about it. We also can use it to compare a car to others by adding more cars on the plot.

Another important feature of my visualization was preattentive processing. Preattentive processing is what allows us to distinguish data quickly (between 50 and 500ms) thanks to a difference of representation. One of the fastest preattentive processing is when you for example decide to put a blue point on a point chart full of red points. This way, your brain will identify in a very short time the blue point, so it basically makes it easier to find specific information.

To add this on my visualization, it is as simple as clicking on a car type in the bar plot. It will then turn all the car’s point colors in red except for the type clicked on which will be then blue. This way, the user can easily identify where are all the cars with a specific feature (here, car maker : Alfa Romeo). It can be very useful when there are too many car types like in the following example with the car makers.

Visualization without preattentive processing
Visualization with Preattentive processing

I personally liked a lot this course. It was very interesting to learn about all these techniques and it was the one with the most real-life use-cases as it is a course mainly focused on design which is basically everywhere in our lives, with a focus here on information visualization.

Sweden’s Universities and Nations

Sweden’s Universities and Nations

When I arrived in Sweden in the end of August, one of the thing that first surprised me is something called « Nations ».

Sweden’s nations are basically groups of people, that everyone can join but nobody is obliged to, with whom you will be able to do a lot of activities prepared by these, like parties and games. Nations seem to be one of the best way to make friends there, to have a lot of fun and it seems to give people a sense of identity as it looks like to me to be big families.

Before the university’s new year starts, they like a lot to party for one or two weeks to enjoy their summer holidays at the fullest.
One way of doing it is to set tents in a green space and to settle there for the time you plan to party which allows you to have your own area to play. The tents also protect you from rain which comes very often in Sweden.

Nations all have their own dress code, it is mainly about having a specific top and a pant of a certain style and a chosen color by the nation where you can hang badges, so basically everyone wears the same clothes. The more you have badges, the more it means you participated in a lot of activities and you are in this Nation for a long time. A friend of me had a pant litteraly full of badges after 4 years.

These nations can have a very big amount of people in it. The nation from the picture was the biggest of all I’ve seen, and I expect their real number to be even bigger as they might not all be there.
In this picture, they were doing some kind of a competitive activity in groups.

Being a part of a nation also allows them to go to specific events or places, there is for example in Växjö a private club that is only for people among nations.

I personally enjoyed seing them having fun and I like a lot the concept of having a big group to which you belong to and in which you spend a lot of time. Even if I couldn’t join a nation, I fortunately could join the Erasmus group which was kind of similar to a nation. It did not have a specific dress code but still we had a lot of activities which includes night parties a bit like the pictures above and also different sports in the university’s gymnasium.

I wish we had something similar in French universities, I’m sure a lot of people would like to participate in something like this, spending time in a big group you can call family.

The five ages of modern man

The five ages of modern man

Collaborative writing, after William Shakespeare’s famous monologue in « As You Like It », Act II Scene VII

At the beginning, the newborn,
Straight from the artificial incubator center, 
One of the hundreds during the past hour,
Ready to live the same life as the others,
The result of too much porn. 

Then as he finishes his class on his tablet,
His mom wants him to explain Hamlet.
But he prefers playing super hero on Fortnite,
And he dreams of becoming a knight.

And then comes the teenager.
In his bag, some paper,
Not only to write but to roll joints,
Phone in hand, liking many pics.

And now, the stressed-out adult,
The one who has no time to himself,
Locked in his routine between kids, work, and mistress,
An exhausted one, with no juice left.

And so comes the end of the journey,
With a heart filled with regret
And forgotten memories of a bygone era.
Nothing left to do, nothing left to see,
Except reminisce about what could’ve been.

by Audran, Damien Da, Damien De, Guillaume, Julien, Ludovic, Nathanaël, Pierre, Raphaël, Sébastien
Spring 2021

How to improve composite task oriented dialogue systems ?

How to improve composite task oriented dialogue systems ?

When a user interacts with an agent (a robot or a program), the dialogue is mainly made up of one or several domains – like a restaurant, a hotel or an attraction –  and one or several tasks – such as searching for a good restaurant which would fit our criteria or to book a hotel. This is a task-oriented dialogue.

A composite dialogue system has a composition of possible dialogue environments. So an agent which includes a composite dialogue system can talk about several domains and do several tasks.

Nowadays, composite task-oriented dialogue systems are efficient to satisfy different tasks about different domains (subjects) such as finding and booking a restaurant or a hotel. However these domains and tasks are defined during the learning step and adding a new domain or task is disabled after this step.

This is due to the representation of the information by the system, let us explain that in this article !

How does a task-oriented dialogue system work ?

Imagine you talk with an agent, the language used in the dialogue is the natural language. To talk with you, the agent needs to translate the natural language into a semantic representation (a language that it understands), it is the ontology. To do so, the dialogue system in the agent uses a Natural Language Understanding module (NLU) and it uses a Natural Language Generation module (NLG) to convert the semantic representation into a natural language.

During the translation, the dialogue system translates each term of the dialogue into dialogue acts, slots and values (Delexicalisation), and puts these on a vector (Vectorisation) to process the request. After the processing, the result is a vector and the DS needs again to transform it into dialogue acts, slots and values (Devectorisation then Lexicalisation).

For example, asking an agent to find the address of an american food restaurant in San Jose will look like that :

So the agent informs the user that there is an American food restaurant in San Jose at the 71th North San Pedro Street.

Improve the vector representation

Currently, the representation of the information in a vector is a master representation, i.e. for one domain (e.g. Restaurant) there are many slots (e.g. food, city, etc) and in one slot there are many values (e.g. Italian, French, Chinese (for food)). This representation is defined during the learning step and it is domain dependent, i.e. adding a value or a slot after the learning phase is disabled.

To improve this representation there is the summary representation. There is no need to know the value but only if there is one or not. Adding a new value is now possible, but the issue of the domain dependency remains.

The representation can be definitely improved with the domain independent parameterization (DIP). It is the same representation as the summary one but the slots in the vector are separated into subvectors so it is now allowed to add values and slots in the representation.

To make the connection with the previous representation : for a domain (e.g. restaurant) there is firstly a vector with the active task (e.g. finding), some requested slots (e.g. food, city and price range) and slots values. Then with the new representation there is one general subvector with some slot independent information – such as the active task, the number of slots, etc – and different slots which do not contain values, but just the information about the presence of the value, 1 if it is present, 0 elseway.

Improve the architecture of the model : architecture based on the DIP representation

The left representation on the illustration is the classic representation of a neural network, i.e. the input module is a vector, the forward module is a layer (matrix) and the output module is also a vector. The issue with this representation is the same as before : it needs to relearn the model to add new values or slots.

That is why, based on the DIP representation, the layer could be composed of as many subblocks as slots in the input module. The information is symmetrical with respect to the diagonal. The aim of this representation is to enable the slot addition and the slot permutation invariance without having to relearn the model.

BONENFANT GUILLAUME

BONENFANT GUILLAUME

Étudiant de la première promotion CMI informatique à Avignon au CERI,
actuellement en master ILSEN 1ère année

Présent au CERI à Avignon depuis la création du CMI informatique, je suis ce cursus pour engranger plus de connaissances sur tout ce qui touche à l’innovation et en apprendre plus sur différents sujets.

J’ai participé au concours « 24h pour entreprendre » qui m’a appris en peu de temps les bases pour monter un business.

A propos de moi :

Ce qui me définit le mieux est « fan du Japon ». J’aime tout de ce pays, de leur culture et leur façon de vivre. Mon objectif est d’aller y travailler un moment, et si je m’y plais, m’y installer.

Les domaines de travail qui me plaisent sont le développement web, la programmation et également la gestion de projet. Ce dernier est celui qui me plaît le plus car j’aime le travail en groupe et j’aimerais bien m’y essayer afin de savoir si cela me convient réellement.

J’ai réalisé un site web permettant d’apprendre les signes de la langue japonaise à l’aide d’exercices. Cette idée m’est venue en voulant apprendre moi-même le japonais, je trouvais l’apprentissage de cette langue beaucoup plus facile et plaisant avec des exercices et n’ait pas trouvé d’équivalent sur internet, j’ai donc fait mon propre site : https://japaniser.fr