Logo: University of Southern California

Events Calendar



Select a calendar:



Filter June Events by Event Type:


SUNMONTUEWEDTHUFRISAT

University Calendar
Events for June

  • PhD Defense - Zahra Nazari

    Wed, Jun 07, 2017 @ 01:00 PM - 02:00 PM

    Computer Science

    University Calendar


    PhD Candidate: Zahra Nazari

    Date: Wed, June 7th
    Time : 1 PM
    Location: KAP134

    Committee :
    Dr. Jonathan Gratch
    Dr. Milind Tambe
    Dr Peter Kim

    Title : Automated Negotiation with Humans


    Negotiation is a crucial skill in personal and organizational interactions. In the last two decades, there has been a growing interest to create agents that can autonomously negotiating with other agents. The focus of this thesis, however, is on creating agents that can negotiate with human opponents. Besides improving on artificial social intelligence, such agents could be used for the purpose of training or assisting human negotiators. A central challenge is to handle the complexity of actual human behavior. When compared with idealized game-theoretic models,
    human negotiations are far richer, both in terms of the nature of information exchanged and the number of factors that inform their decision-making.

    We consider a negotiation task that is simple, yet general enough to drive agent-human research, and
    analyze an extensive data set of transcribed human negotiation on such tasks.
    Based on human behavior in this task, and the previous research on human negotiations, we propose a new framework to structure the design of agents that negotiate with people. We address two main decision problems inspired by this framework: modeling and influencing the opponent. Three techniques are proposed to model an opponent's preferences and character (e.g. honesty and personality traits) and a misrepresentation technique is then used to influence the opponent and gain better profit. The proposed techniques are then implemented in automatic web-based agents. We ran a number of negotiations between these agents and humans recruited on Amazon Mechanical Turk. The resulting data show that the agents can perform these strategies successfully when negotiating with human counterparts and give us valuable insight about the behavior of humans when negotiating with an agent.

    Location: Kaprielian Hall (KAP) - 134

    Audiences: Everyone Is Invited

    Posted By: Lizsl De Leon

    OutlookiCal
  • PhD Defense - Rose Yu

    Wed, Jun 07, 2017 @ 01:30 PM - 03:30 PM

    Computer Science

    University Calendar


    PhD Candidate: Rose Yu

    Date: June 7, 2017
    Time: 1:30-3:30pm
    Location: SAL 213

    Committee:
    Yan Liu
    Cyrus Shahabi
    Mahdi Soltanolkotabi (outside member)


    Title:
    Tensor learning for Large-Scale Spatiotemporal Analysis

    Abstract:
    Spatiotemporal data is ubiquitous in our daily life, including climate, transportation,
    and social media. Today, data is being collected at an unprecedented scale.
    Yesterdays concepts and tools are insufficient to serve tomorrow's data-driven
    decision makers. Particularly, spatiotemporal data often demonstrates complex
    dependency structures and is of high dimensionality. This requires new machine
    learning algorithms that can handle highly correlated samples, perform efficient
    dimension reduction, and generate structured predictions.

    In this talk, I will present tensor methods, a scalable framework for capturing
    high-order structures in spatiotemporal data. I will demonstrate how to learn from
    spatiotemporal data efficiently in both offline and online setting. I will also show
    interesting discoveries by our methods in climate and social media applications.

    Location: 213

    Audiences: Everyone Is Invited

    Posted By: Lizsl De Leon

    OutlookiCal
  • PhD Defense - Simon Woo

    Mon, Jun 12, 2017 @ 09:00 AM - 11:00 AM

    Computer Science

    University Calendar


    PhD Candidate: Simon Woo
    Date: June 12, 2017
    Time: 9:00am-11:00am
    Location: SAL 322
    Committee:
    Jelena Mirkovic (Adviser)

    Ron Artstein

    Kevin Knight

    Elsi Kaiser (outside member)

    Title: MEMORABLE, SECURE, AND USABLE AUTHENTICATION SECRETS


    Abstract:
    Textual passwords are widely used for user authentication, but they are often difficult for a user to recall, and easily cracked by automated programs, and heavily re-used. Weak or reused passwords are guilty for many contemporary security breaches. Hence, it is critical to study both how users choose and reuse passwords, and the reasons that they adopt unsafe practices. In this thesis, I first examine the reasons why people create weak passwords and reuse these over multiple accounts. My research complements the body of existing works by studying the semantic structure, strength and reuse of real passwords, as well as conscious and unconscious causes of unsafe practices, using a test group population of 50 participants. Significant reuse and weak passwords clearly demonstrate the need for alternative authentication methods that are more memorable, secure, and less reused. My next three key thesis topics focus on developing novel authentication mechanisms that can directly improve current approaches. The first approach, "Life-Experience Passwords (LEPs)." uses a person's prior life experience as information to generate more memorable and secure authentication questions. We show that LEPs significantly raise the level of memorability and security compared to existing passwords and security questions. My second approach constructs more memorable and more secure passphrases through the novel use of mnemonics - multi-letter abbreviations of passphrases (MNPass), made of the first letters of each word in a passphrase. I apply mnemonics when generating and authenticating passphrases and show that the mnemonics-based approach improved recall compared to randomly generated passphrases and enhanced strength compared to user-selected passphrases. My last work explores password creation with semantic feedback (GuidedPass). I analyze user-input passwords and provide real-time, specific suggestions for improvement based on their existing semantic structure. GuidedPass passwords are 10^4 to 10^7 times stronger and as memorable as user initial passwords. GuidedPass passwords are also 100 times stronger and 1.2 times more memorable than passwords created with only password-meter feedback.

    Bio:
    Simon Woo is a Ph.D. candidate advised by Prof. Jelena Mirkovic. His current research focuses on improving user authentication, and understanding human factors in cybersecurity to better design secure systems.

    Location: 322

    Audiences: Everyone Is Invited

    Posted By: Lizsl De Leon

    OutlookiCal