Complexity Science » ��TV Annual Retreat Projects

Fungy Fun with Befunge-93, a two-dimensional programming language

Edoardo Barp — Sat, 28 Apr 2018 11:38:25 GMT

Last but not least addition to the WARPs, I present you: BEFUNGE-93!

Abstract

Befunge-93 is a programming language in which programs are not written in lines of code, but grid based and two dimensional, and where each cell contains a single instruction, and execution can proceed in any cardinal direction across this grid -- not just left-to-right, but also right-to-left, top-to-bottom, and bottom-to-top. Additionally, note that the program is self-modifying, that is the grid can be edited through special characters as the program is running.

Aim and objectives

In this project, we aim to first discover and have some fun with the language, then try to implement simple models, and additionally modify the interpreter to add the functionality to call other files (so that one could potentially separate functions into files). By the end of the project, we aim to be world experts in fungeoid languages and in self-modifying code programming.

Of interest to

Anyone who thought "this sounds fun" as they were reading the above chapters

Resources necessary

Interpreter - https://github.com/catseye/Befunge-93

Further reading

https://en.wikipedia.org/wiki/Befunge

Re: Fun with Generative Models

Iliana Peneva — Wed, 25 Apr 2018 16:30:31 GMT

Here are the first few datasets: https://drive.google.com/drive/folders/1OSWWJDyPm305_GuUmg_64rvyaoHxyHCq?usp=sharing

These include: cats, dogs, Pokemon, DnD spells (5th edition), wrestlers' names (thanks to Jack :-) ) and rom coms. I will upload the talk abstracts from previous annual retreats and cake recipes later this week.

Re: Functional Programming Madness

Jonathan Skipp — Fri, 13 Apr 2018 13:12:56 GMT

https://xkcd.com/1270/

Mouseover text: "Functional programming combines the flexibility and power of abstract mathematics with the intuitive clarity of abstract mathematics."

Fun with Generative Models

Iliana Peneva — Wed, 04 Apr 2018 18:28:37 GMT

Fun with Generative Models

Organised by Iliana Peneva

Abstract

Neural networks have been used for almost everything: to train a robot how to play football; to generate April Fools jokes/names of metal bands/traditional Irish music/candy heart messages, to distinguish between chihuahua and muffin, to generate cats from drawings. Inspired by Rob's entertaining WARP from 2 years ago and by Janelle Shane's brilliant blog, my proposed project aims to add to this list of fun NNs applications by generating new dog/cat breeds, DnD spells, Pokemons, cake recipes, talk abstracts, news items.

We can decide what to generate at the retreat (there can be people working on different topics)
Suggestions are welcome for other topics :-)

Aims and Objectives

learn more about different types of neural networks, their capabilities and limitations
get experience using neural networks in Python/R

Of Interest to

Anyone interested in deep learning (or likes cats/dogs/cakes/DnD/Pokemon)

Resources Necessary
The datasets will be collected and made available before the retreat.

if using Python - Keras and Tensorflow; if using R - neuralnet, or Keras and Tensorflow for R

References
(most relevant)
Andrey Karpathy's excellent blogpost with examples: http://karpathy.github.io/2015/05/21/rnn-effectiveness/
Christopher Olah's blogpost: http://colah.github.io/posts/2015-08-Understanding-LSTMs/

(for the keener)
Deep Leaning book by Ian Godfellow, Yoshua Bengio, Aaron Courville: http://www.deeplearningbook.org/contents/rnn.html

Re: Functional Programming Madness

Jack Binysh — Fri, 16 Mar 2018 15:29:17 GMT

https://insights.stackoverflow.com/survey/2018/

Haskells super hot right now too, a whole 6% of devs want to use it

Trait-based phylogenetics and the evolution of EVERYTHING(!!!)

Joe Hilton — Fri, 16 Mar 2018 12:09:32 GMT

Trait-based phylogenetics and the evolution of EVERYTHING(!!!)

Organised by - Joe Hilton

Abstract -

Phylogenetics is the study of the evolutionary relationship between groups of organisms. By comparing the phenomes and/or genomes of different species we can infer an evolutionary tree which visualises their evolutionary history. Phylogenetic methods are also used outside of biology to construct family trees for other kinds of entities which have some sort of evolutionary relationship, most notably languages. This project is an introduction to trait-based (in biological terms phenome-level) phylogentics using the Matlab package Traitlab. Traitlab was primarily developed to construct language family trees, but can be used to infer relationships between any set of entities for which we can provide a list of traits and their presence/absence. My plan is to get some suitable datasets together (possibly before the retreat) and then let Traitlab loose on them, hopefully figuring out how it works along the way. Since these datasets can come from just about anywhere, we can use Traitlab to generate "family trees" for lots of weird non-biological stuff, possibly throwing any sort of scientific rigour out the window. Some examples I've thought of are:

Metal subgenres (traits could include speed, growliness, and spookiness)
Pokemon species (traits could include type, number of evolutionary stages, standard phenomic stuff from biology)
Types of sandwich (fillings, type of bread, country of origin etc)

If you're interested in the project, try to come up with some ideas for datasets so this isn't just "let's classify stuff Joe likes". If all goes well we will be experts in phylogenetics by the end of the retreat. If not, we will at least have some nice pictures of trees that we can show to people.

Aims and Objectives -

Get to grips with some of the basic concepts of phylogenetics
Get some experience using statistical inference and MCMC in a practical setting
Misuse scientific concepts and experience a thrill of naughtiness
Obtain biological proof that the sprites for Butterfree and Venomoth are the wrong way round

Of Interest to -

Mathematical biologists, people who like MCMC/inference, linguistics geeks, geeks in general

Resources Necessary -

Traitlab package for Matlab, available here: https://sites.google.com/site/traitlab/
Between now and the retreat it's possible that I'll find other/better packages, in which case I'll add them to the thread.
Fun datasets/ideas for datasets. Since we possibly/probably/definitely won't have wi-fi, it'll be useful to either have data or a useful way to construct data before the project starts.

References -

G. K. Nicholls, R. J. Ryder and D. Welch, TraitLab: a MatLab Package for Fitting and Simulating Binary Trait-Like Data, Technical report

Recommended background reading for those interested in the project -

https://en.wikipedia.org/wiki/Phylogenetics
https://www.nature.com/scitable/topicpage/trait-evolution-on-a-phylogenetic-tree-relatedness-41936
Not super relevant but a cool piece of research using phylogenetics: http://rsos.royalsocietypublishing.org/content/3/1/150645

Designing Games to Convey Scientific Concepts

Annika Heike Stechemesser — Thu, 15 Mar 2018 15:51:14 GMT

Designing Games to Convey Scientific Concepts

Organised by -

Annika Stechemesser

Abstract -

Working in an interdisciplinary environment, communicating our research to a non specialist audience is one of the daily challenges. We might not only want to explain our thoughts and results to fellow university researches from other disciplines, but also to partners in industry and to the wider public. Thinking about ways in which we can break our work down into smaller, approchable steps that relate to other peoples experiences and interests is therefore cruical. Ideally, we don't just want others to just understand, but to feel inspired and keen to learn more.

This warp projects invites you to think about how you can communicate your research in a way that makes people with no prerequisites feel excited about it. One method of conveying simplified scientific concepts to the public is through games.

Some examples:

(1) In the Village (http://www.tiltfactor.org/game/in-the-village/)

In the card game In the Village, players must cooperate and use limited resources to ensure their survival against the threat of malaria. In the Village was originally designed as a challenge: Could a game demonstrate to players the basic ideas behind malaria prevention, and foster a reliance on community-based health care solutions?
The game is designed for five players (or, in smaller groups, one player may take on more than one character). Each player tries to keep his or her character(s) alive through nights of mosquito attacks, with the caveat that if even one player dies, the game is over.

While the game’s five players are given the agency to act as individuals, the victory and loss conditions apply to all players equally, suggesting open and selfless cooperation as the optimal strategy. Despite this, playtesting found that many players would guard their own resources and try to protect their perceived individual interests. Winning the game requires overcoming this instinct towards selfishness. Players end up sharing resources to make sure the group survives. The game offers a compelling model for community caregiving while offering fun, engaging gameplay.

(2) Zombiepox (http://www.tiltfactor.org/game/zombiepox/)

ZOMBIEPOX® has reached your town and turned two people into full-blown zombies! ZOMBIEPOX spreads as the zombies wander through town biting people! Help the humans escape the wrath of the flesh-eating zombies by vaccinating them against ZOMBIEPOX. Cure those in the most dire situations and protect the vulnerable who cannot be immunized. We don’t want zombie babies! Win the game if the deadly ZOMBIEPOX can no longer spread. Lose if too many people become full-blown zombies and the town is overrun!

(3) The Trap (https://copyrightliteracy.org/resources/the-publishing-trap/) (This one was even designed to educate us...)

The Publishing Trap is a game about research dissemination and scholarly communication in Higher Education. The game follows the academic career of four characters who at each stage in their career, from PhD submission, through to Professorship, are presented with a series of scenarios about which they have to make choices. The characters make decisions about how to disseminate their research at conferences, in academic journals and in monographs or textbooks. Ultimately the game helps researchers to understand how money, intellectual property rights, and both open and closed publishing models affect the dissemination and impact of their research. Through playing the game in teams, players get to discuss the impact of each character’s choices. The game ends at the end of the character’s life, when players sees the consequences of the choices they have made in terms of money, knowledge and impact.

In this WARP project, we will have a go at designing our own games to explain our research!

More inspiration: http://www.tiltfactor.org/

Aims and Objectives -

The final aim is down to the wishes of the people participating (and also the number of people participating), I could see two things working well:

1. everyone gives designing a game based on their own research a go, possible stepts could be:

come up with a general concept (WHAT do I want to communicate, to WHOM, in which WAY)
create a set of rules, identify possible resources
start to implement the game, either digitally (for example with PYGAME or whichever suitable language you prefer!) or as a tabletop game (I will bring some crafts stuff)

2. People team up and we focus on one topic, designing one game together. This would probably result into a tabletop game as it is easier to work on with multiple people at the same time. The steps stay essentially the same.

Of Interest to -

Anyone interested in thinking about how to break down their research for others to understand more easily, anyone having fun being creative, anyone liking games.

Resources Necessary -

If you are keen to implement something digitally maybe download PYGAME or whichever tool you prefer to write games in. If you want to get really far in your game design in these 6 hours maybe go for a head start and watch some tutorials etc.

References -

https://www.nature.com/naturejobs/2017/170720/pdf/nj7663-369a.pdf

http://news.mit.edu/2011/vanished-smithsonian-0415

http://blogs.lse.ac.uk/impactofsocialsciences/2017/10/28/the-publishing-trap-a-game-of-scholarly-communication/

Sentiment analysis using Twitter data

Sophie Meakin — Wed, 14 Mar 2018 23:27:07 GMT

Sentiment analysis using Twitter data

Organised by Sophie Meakin

Abstract

Sentiment analysis uses natural language processing (NLP) to identify and quantify an individuals opinion on a topic. Twitter is one example of where individuals express their opinions about many different subjects. My proposed project is to collect data on a certain topic and then analyse the content, for example:

(1) PhD sentiment analysis: The #PhDChat hashtag was originally created as a way for PhD students in the UK to hold weekly discussions, but it has grown into much more than that and is now used by PhD students and others around the world to talk about the challenges and successes of studying for a PhD, ask for advice, and provide support. We can collect tweets containing the #PhDChat hashtag and analyse the tweets using Python libraries for natural language processing/ sentiment analysis.

(2) We could collect data on anything, suggestions are very welcome! For example:

Tweets from Donald Trump: https://dev.to/rodolfoferro/sentiment-analysis-on-trumpss-tweets-using-python-
Brexit: http://bruegel.org/2016/11/tweeting-brexit-narrative-building-and-sentiment-analysis/

Aims and Objectives

(1) To understand the sentiment of Twitter users using the #PhDChat hashtag.

Of Interest to

Anyone interested in natural language processing, novel uses of social media data, or anyone who has ever wondered how much other people are enjoying their PhDs.

Resources Necessary

Data will be collected and made available before the retreat. Updates on other resources required to follow.

References

The Structure and Characteristics of #PhDChat, an Emergent Online Social Network: https://www.youtube.com/watch?v=64uSxFeeV5s

Re: Pommerman – Multi-Agent Learning Competition

Henry Charlesworth — Mon, 05 Mar 2018 11:59:09 GMT

OK cool, glad to hear it! I'm also not at all an expert in reinforcement learning but it's an area that I've learning about recently and getting pretty interested in. I've had a go at implementing some Deep Q networks and simple policy gradient algorithms (but my understanding is that these don't actually work all that well in multi-agent environments) so we probably need to do something a bit different. Still I was thinking tensorflow because probably training a neural network of some kind is the only approach I can think of at the moment (e.g. maybe trying to learn some more about the competitive self-play methods they've been using at OpenAI). I'm definitely open to suggestions though!

Re: Pommerman – Multi-Agent Learning Competition

Edoardo Barp — Sun, 04 Mar 2018 14:19:01 GMT

This is definitely somethign I'm interested it!

I've never implemented or tested any reinforcement learning, although I now have some understanding (thanks to a module) of basic reinforcement learning methods (Q-learning, MC tree search..).

How were you thinking of using tensorflow? I've only used it for neural networks, which I thought was the only thing it did. Were you planning on training a NN using episode learning for instance?

Great project anyway!

Functional Programming Madness

Jack Binysh — Sat, 03 Mar 2018 13:49:17 GMT

Functional Programming Madness

Organised by: Jack Binysh

Abstract : Functional programming languages are a class of languages (Haskell, Lisp, bits of Python, etc.) which employ an approach to programming different to imperative, C style programming - they focus on describing what a function is, rather than how it is to be implemented. Eg, here's a quicksort in Haskell:

quickSort [] = []
quickSort (x:xs) = quickSort (filter (<x) xs)
++ [x] ++
quickSort (filter (>=x) xs)

... it looks like top drawer nerd madness.

I know nothing about functional programming but thought it might be fun to learn some Haskell.

Aims and Objectives:

Learn some Haskell, code something up in it. Reach programming Nirvana.

Of Interest to:

People who want their minds blown.

Resources Necessary:

Haskell

https://wiki.haskell.org/Haskell

References:

https://xkcd.com/224/

https://xkcd.com/297/

https://wiki.haskell.org/Why_Haskell_matters

http://learnyouahaskell.com/

Google Landmark Retrieval Challenge

Bhavan Chahal — Wed, 28 Feb 2018 22:41:48 GMT

Google Landmark Retrieval Challenge

Organised by: Bhavan Chahal & Chris Norman (two groups, same problem)

Abstract:

Image retrieval is a fundamental problem in computer vision: given a query image, can you find similar images in a large database? This is especially important for query images containing landmarks, which accounts for a large portion of what people like to photograph.

In this Kaggle competition, competitors are given query images and, for each query, are expected to retrieve all database images containing the same landmarks (if any).

The new dataset is the largest worldwide dataset for image retrieval research, comprising more than a million images of 15K unique landmarks. We hope that this release will accelerate progress in this important research problem.

If we decide to submit, submissions are evaluated according to mean Average Precision, and the prizes are 1st place - $ 1,250, 2nd place - $ 750, and 3rd place - $ 500. The deadline is May 15, 2018, a few weeks after the retreat.

Aim:

Given an image, can you find all of the same landmarks in a dataset? We would like to have two groups approaching the problem differently and see which performs best!

Objectives:

Analyse the data and format as required
Attempt the Kaggle competition
Have fun with data science!

Of Interest to:

Google, data scientists, and those interested in vision tasks

Resources Necessary:

Competition data (Available here: https://www.kaggle.com/c/landmark-retrieval-challenge/data)
R or Python and relevant packages

References:

Kaggle competition page: https://www.kaggle.com/c/landmark-retrieval-challenge
H. Noh, A. Araujo, J. Sim, T. Weyand, B. Han, "Large-Scale Image Retrieval with Attentive Deep Local Features", Proc. ICCV'17

Pommerman – Multi-Agent Learning Competition

Henry Charlesworth — Fri, 23 Feb 2018 21:54:32 GMT

Pommerman – Multi-Agent Learning Competition

Organised by -
Henry Charlesworth

Abstract -
Pommerman is an implementation of the SNES game “Super Bomberman”designed such that it can be used as a playground for multi-agent learning. This is a very simple four player game with two different game modes (FFA and Team 2v2) but is made complicated by the fact that it is adversarial and you are competing against other agents. The idea of the Pommerman project is that people from around the world will try and train agents to play this game, submit their entries and then the organizers will host competitions where these all compete against each other.

Aims and Objectives -
Realistically it will be difficult to create a highly competitive agent within the amount of time we have but hopefully it will be possible implement something simple that can at least play the game sensibly and then to brainstorm ideas that we could try to implement afterwards if people are still interested.

Of Interest to -
People interested in machine learning (particularly reinforcement learning) and people who like old SNES games. I am mainly interested in doing this because I’d like a chance to try and implement some reinforcement learning algorithms.

Resources Necessary -
Agents can be developed using any framework essentially (the only requirement is that you submit a Docker container). I was planning to use tensorflow but am open to suggestions from people who know a lot more about machine learning than I do.

References -
https://github.com/MultiAgentLearning/playground
for the very keen: https://github.com/LantaoYu/MARL-Papers

Modelling the Emergence and Collapse of Eusociality

Robert Gowers — Thu, 22 Feb 2018 11:18:08 GMT

Modelling the Emergence and Collapse of Eusociality

Organised by -

Robert Gowers

Abstract -

Why and how eusociality evolved in certain animals has been a puzzling question for many years. Many different hypotheses have been advanced to explain this, but later studies have shown that simple explanations are insufficient to account for all different species types. Mathematical models have been devised (see the third reference) that detail how different evolutionary traits allow for eusociality. However, many questions still remain.

Aims and Objectives -

This project intends to create a simple model of a population with some genetic traits to see not only how eusociality can emerge, but also how it can collapse. How robust is a eusocial population to perturbations? Can a species simply revert back to solitary behaviour after being eusocial, or will it simply die off?

More specific objectives and information will be added closer to the time of the retreat.

Of Interest to -

People who like population models, people who like genetic models, people who like ants.

Resources Necessary -

No special software should be required.

References -

Eusociality, origin and consequences, http://www.pnas.org/content/pnas/102/38/13367.full.pdf

The evolution of eusociality, https://www.nature.com/articles/nature09205

A unified model of Hymenopteran preadaptations that trigger the evolutionary transition to eusociality, https://www.nature.com/articles/ncomms15920

Sexually Transmitted Infections in a Brave New World

Trystan Leng — Mon, 19 Feb 2018 09:30:18 GMT

Sexually Transmitted Infections in a Brave New World

Organised by - Trystan Leng

Abstract -

In his 1931 novel Brave New World, Aldous Huxley imagines a dystopian society where social norms are radically different from our own. The spread of infections, and in particular the spread of sexually transmitted infections (STIs), can be thought of as processes on networks - hence a change in social norms, by changing the structure of these networks, has the potential to alter both the spread and control of STIs drastically. This project will create appropriate models to explore the effect the differing social norms between our world and Huxley's Brave New World have on the spread and control of STIs. Doing so provides an insight into which aspects of sexual networks must be explicitly modelled, and more importantly, which can we neglect without affecting the results of our models.

The project may use pair-formation models or stochastic simulations of networks, depending upon preliminary discussions within the group. Pair-formation models have the advantage of being in general deterministic, and hence more tractable, but to assess the impact of some control measures (such as contact tracing) a simulation of the full model might be necessary. The project may be partitioned along lines of different modeling methods, different disease dynamics, or different control interventions - to be decided within the group at the retreat.

Aims and Objectives -

Aim-to explore the impact differing social norms has on the spread and control of STIs, to better understand which features of networks should be explicitly modelled for the spread of infectious diseases.

Objectives -

Determine the ways in which social and sexual norms differ between our world and in a Brave New World, and decide which of these to focus on.
Construct appropriate models of STI spread under differing social norms.
Examine how the spread and control of STIs differ between the constructed models, in particular when parameters are adjusted to achieve the same population level quantities across models (such as prevalence).

Of Interest to - Those interested in epidemiology and processes on networks.

Resources Necessary -

Matlab code for base pair-formation model provided by Trystan
download Rstisim for stochastic simulation of STIs https://blog.nus.edu.sg/roellin/rstisim/

References -

An introduction to the themes of Brave New World: https://www.shmoop.com/brave-new-world/
Kretzschmar and Heijne's primer on pair formation models for STIs: https://www.sciencedirect.com/science/article/pii/S2468042717300106

WARPS Project Template -

Trystan Leng — Mon, 19 Feb 2018 09:21:18 GMT

WARPS Project Title

Organised by -

Yourself - you are responsible for organising the project and guiding your group towards its aims and objectives

Abstract -

An introduction to and motivation behind the project. Abstracts preferably should be aimed at a non-specialist audience - i.e. aimed at those in the MathSys CDT but who are not necessarily experts in the given field. Once introduced, an explanation of the structure of the project is recommended. Projects should aim to be fun, challenging and informative, with achievable objectives for the end of the retreat.

Aims and Objectives -

The overall aims and the specific objectives of the project - what do you hope to have achieved by the end of the retreat?

Of Interest to -

Name fields or skill-sets to whom the project may be of specific interest to.

Resources Necessary -

Any resources of software necessary for group members to obtain prior to the retreat.

References -

Recommended background reading for those interested in the project.

Complexity Science » ����TV Annual Retreat Projects

Fungy Fun with Befunge-93, a two-dimensional programming language

Re: Fun with Generative Models

Re: Functional Programming Madness

Fun with Generative Models

Re: Functional Programming Madness

Trait-based phylogenetics and the evolution of EVERYTHING(!!!)

Designing Games to Convey Scientific Concepts

Sentiment analysis using Twitter data

Re: Pommerman – Multi-Agent Learning Competition

Re: Pommerman – Multi-Agent Learning Competition

Functional Programming Madness

Google Landmark Retrieval Challenge

Pommerman – Multi-Agent Learning Competition

Modelling the Emergence and Collapse of Eusociality

Sexually Transmitted Infections in a Brave New World

WARPS Project Template -

Complexity Science » ��TV Annual Retreat Projects