Upcoming Seminar

Date: February 18 (next Thursday)
Time: 1:00-2:00pm
Location: McConnell Room 320
 
Speaker: Pierre-Luc Bacon
Title: Time, bounded rationality and representations in reinforcement learning
 

Abstract:
In this talk, I will tell about some of the research that I’ve been conducting with Prof. Doina Precup over the last couple years.  The main topic will be the problem of “temporal representation learning” in reinforcement learning. Reinforcement learning is an Artificial Intelligence approach to the problem of sequential decision making, in a world full of uncertainty and under limited computational capacity. As for “representation learning”,  it refers to the problem of autonomously finding and  expressing knowledge within a particular reasoning structure while also improving it over time through experience. I will develop  these ideas through the notion of “bounded rationality” and present some recent mathematical tools that we developed to tackle  this problem. In a sense, the title of this talk also reflects my experience through PhD: a journey to improve and refine my own subjective understanding of the world (and of myself) under limited capacity. Just as for our reinforcement learning agents, I had to  embrace the stochasticity of life while leveraging its regularities. I will try to share both sides of the story: how the research results came about, and how I became more of a researcher over time.

Seminar 4 : Code Fragment Summarization

The Computer Science Graduate Society is pleased to present a talk by Annie Ying on February 11, Thursday. To help us know how much food to order, please fill this Google form if you plan to attend: http://goo.gl/forms/HLEmeB3DZR.

—————–

Date: February 11

Time: 1:00-2:00pm

Location: McConnell Room 320

Speaker: Annie Ying

Title: Code Fragment Summarization

Abstract:
Code fragments are an important resource for understanding the Application Programming Interface (API) of software libraries. Many usage scenarios for code fragments require them to be distilled to their essence: for example, when serving as cues to longer documents, for reminding developers of a previously known idiom, or for displaying search results. This dissertation reports on research on shortening, or summarizing, code fragments and makes three main contributions: a set of lessons learned from a case study on a supervised machine learning approach to the generating code fragment summaries; an empirically grounded catalog of source code summarization practices; and the design, implementation and evaluation of a novel optimization-based summarization technique for code fragments.

Afternoon Tea Every Thursday

We are starting “Afternoon tea” events for graduate students and faculty members.  Every Thursday from 3-4 pm the lounge will be reserved for us.  CSGS will provide the tea and cookies! We will also buy a new tea boiler for the lounge. (yaaaay!).  Please come with your own mug.  We hope this will be a good way for all of us to connect and to share our research and non-research interests! There will be plenty of chalks available as well (maybe colourful ones too)!

We start this week!

Seminar 3 : Marrying Reinforcement Learning and Deep Learning

The Computer Science Graduate Society is pleased to present the first talk of the winter semester by Emmanuel Bengio on 15th January. To help us know how much food to order, please fill this Google form if you plan to attend: http://goo.gl/forms/9lVPNT5tyV

—————–

Date: January 15
Time: 13:00-14:00
Location: McConnell Room 103

Speaker: Emmanuel Bengio
Title: Marrying Reinforcement Learning and Deep Learning

Abstract: In the last few years Machine Learning has boosted in popularity thanks to the outrageously successful applications of Deep Learning, in many areas including speech recognition, fraud detection, advertisement, recommendation systems, and, probably most famously, computer vision.
Taking advantage of these techniques, we have also very recently seen successful use of Deep Learning as a tool inside of Reinforcement Learning tasks, such as DeepMind’s popular Atari model. A new approach, which is the main concern of my research, consists in studying the opposite direction: using Reinforcement Learning to augment Deep Learning models. This takes the form of conditional computing, visual attention, memory mechanisms, and much more, which I will discuss during this talk.

Seminar 2 : Efficient Collaborations with Trust-Aware Robots

The Computer Science Graduate Society is pleased to present a talk by Anqi Xu on 2nd December (next Wednesday). To help us know how much food to order, please fill this Google form if you plan to attend: http://goo.gl/forms/hkof94CtJC.

Date: December 2 (next Wednesday)

Time: 12:00-13:00

Location: McConnell Room 103

Speaker: Anqi Xu

Title: Efficient Collaborations with Trust-Aware Robots

Abstract:

In this work, we give autonomous robot agents the ability to infer their human collaborator’s changing trust states, and consider how this signal can be capitalized to improve the efficiency of human-robot teams. This trust-aware robot framework incorporates advances in online human-robot trust modeling and interactive behavior adaptation for autonomous agents. We build upon these components by introducing the novel formulation of trust-induced conservative control. This enables the robot agent to momentarily alter its behaviors in response to the human’s trust losses, as an active means to mend damage to the team relationship. We present two end-to-end instantiations of trust-aware robots for distinct task domains of aerial terrain coverage and interactive autonomous driving. Our empirical assessments comprise of a large-scale controlled study, as well as field evaluations with a smart car platform. These assessments quantitatively demonstrate the diverse efficiency gains of trust-aware robots.

Financial Statement

Transparency is next to Godliness!! Here we tell you, most cherished CS grads, about the ways in which we are spending your money. Don’t hesitate to contact us if you feel we are over- or  under-spending on some criteria!.

DateAmountExpenseDetailLocation
11/09/2015$191.80Seminar1st seminar food and drinks for grad studentsMcConnell 320
11/06/2015$37.56Seminar1st seminar hoodie gift and gift bag for speakerMcConnell 320
11/30/2015$76.85Seminar2nd seminar hoodie gift and gift bag for speakerMcConnell 320
12/02/2015$170.40Seminar2nd seminar foodMcConnell 103
01/13/2016$438.41Seminar10 hoodies for the upcoming 10 seminar+1.5 e-transfer feeMcConnell 103
01/13/2016$190.05Seminar3rd seminar foodMcConnell 103
02/08/2016$246.61Board Games NightFood, drinks and gamesThomson House

Our first Survey hooray!

Following are the response charts :

Q1: We will cater lunch from your favorite restaurant(s) for our seminar lunches. Rank the following  choices from 1 to 5, your most favorite being 1 (google the name of the restaurant to learn more about it) :

 Screen Shot 2015-11-09 at 5.06.21 PM
Q2: What day of the week would you prefer the seminars to be held? Rank the days from 1 to 5:
 Screen Shot 2015-11-09 at 5.31.45 PM
Q3: What time should seminars be held ? Pick the time that most suits your schedule:
 Screen Shot 2015-11-09 at 5.32.31 PM
Q4: The names of CS grads attending each seminar will be collected for a raffle (lottery) to be held a year from now (Fall 2016), where one name will be drawn to be the winner of a tablet  (hence the more seminars a student attends the more chance s/he will win the tablet). Do you like this idea ?
Screen Shot 2015-11-09 at 5.33.10 PM

Seminar 1 : RNA Sequence Design

newanigif

Speaker: Vladimir Reinharz
Title: RNA Sequence Design

Abstract:
RNAs are biomolecules that can be seen as words in {ACGU}+. An interesting mathematical representation of the structure as a dot bracket sequence [e.g. (((…))) ] has been shown to have some ressemblance with reality. Having an energy model derived from experiments, the ensemble of structures given a sequence has been completely characterized in the 70s (in this model). At the opposite, given a structure, knowing if it exists a sequence with an affinity higher than some threshold  is believed to be NP-hard. I developed an algorithm based on the inside-outside to, given a structure, do a global stochastic Boltzmann sampling of sequences in linear time. I will also talk of a few potential applications in drug design and bioengineering. It has been implemented in a tool called incaRNAtion

http://dx.doi.org/10.1093/bioinformatics/btt217