Select a calendar:
Filter May Events by Event Type:
SUNMONTUEWEDTHUFRISAT
University Calendar
Events for May
-
Study Day
Tue, May 01, 2018
Viterbi School of Engineering Student Affairs
University Calendar
Audiences: Everyone Is Invited
Contact: Sheryl Koutsis
-
PhD Defense - Marjan Ghazvininejad
Mon, May 07, 2018 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Candidate: Marjan Ghazvininejad
Committee: Kevin Knight (chair), Morteza Dehghani, Jonathan May
Title: Neural Creative Language Generation
Time & place: Monday, May 7th, 10 am, GFS 204
Abstract: Natural language generation (NLG) is a well-studied and still very challenging field in natural language processing. One of the less studied NLG tasks is the generation of creative texts such as jokes, puns, or poems. Multiple reasons contribute to the difficulty of research in this area. First, no immediate application exists for creative language generation. This has made the research on creative NLG extremely diverse, having different goals, assumptions, and constraints. Second, no quantitative measure exists for creative NLG tasks. Consequently, it is often difficult to tune the parameters of creative generation models and drive improvements to these systems. Lack of a quantitative metric and the absence of a well-defined immediate application makes comparing different methods and finding the state of the art an almost impossible task in this area. Finally, rule-based systems for creative language generation are not yet combined with deep learning methods. Rule-based systems are powerful in capturing human knowledge, but it is often too time-consuming to present all the required knowledge in rules. On the other hand, deep learning models can automatically extract knowledge from the data, but they often miss out some essential knowledge that can be easily captured in rule-based systems.
In this work, we address these challenges for poetry generation, which is one of the main areas of creative language generation. We introduce password poems as a new application for poetry generation. These passwords are highly secure, and we show that they are easier to recall and preferable compared to passwords created by other methods that guarantee the same level of security. Furthermore, we combine finite-state machinery with deep learning models in a system for generating poems for any given topic. We introduce a quantitative metric for evaluating the generated poems and build the first interactive poetry generation system that enables users to revise system generated poems by adjusting style configuration settings like alliteration, concreteness and the sentiment of the poem. The system interface also allows users to rate the quality of the poem. We collect users' rating for poems with various style settings and use them to automatically tune the system style parameters. In order to improve the coherence of generated poems, we introduce a method to borrow ideas from existing human literature and build a poetry translation system. We study how poetry translation is different from translation of non-creative texts by measuring the language variation added during the translation process. We show that humans translate poems much more freely compared to general texts. Based on this observation, we build a machine translation system specifically for translating poetry which uses language variation in the translation process to generate rhythmic and rhyming translations.
Location: 204
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Shuyang Gao
Mon, May 07, 2018 @ 12:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Mutual Information Estimation and Its Applications to Machine Learning
PhD Candidate: Shuyang Gao
Date: May 7
Time: 12pm
Location: SOS B37
Committee: Aram Galstyan, Greg Verg Steeg, Ilias Diakonikolas, Aiichiro Nakano, Roger Ghanem
Abstract:
Mutual information (MI) has been successfully applied to a wide variety of domains due to its remarkable property to measure dependencies between random variables. Despite its popularity and wide spread usage, a common unavoidable problem of mutual information is its estimation. In this thesis, we demonstrate that a popular class of nonparametric MI estimators based on k-nearest-neighbor graphs requires number of samples that scales exponentially with the true MI. Consequently, accurate estimation of MI between strongly dependent variables is possible only for prohibitively large sample size. This important yet overlooked shortcoming of the existing estimators is due to their implicit reliance on local uniformity of the underlying joint distribution. As a result, my thesis proposes two new estimation strategies to address this issue. The new estimators are robust to local non-uniformity, works well with limited data, and is able to capture relationship strengths over many orders of magnitude than the existing k-nearest-neighbor methods.
Modern data mining and machine learning presents us with problems which may contain thousands of variables and we need to identify only the most promising strong relationships. Therefore, caution must be taken when applying mutual information to such real-world scenarios. By taking these concerns into account, my thesis then demonstrates the practical applicability of mutual information on several tasks, and our contributions include
i) an information-theoretic framework for measuring stylistic coordination in dialogues. The proposed measure has a simple predictive interpretation and can account for various confounding factors through proper conditioning ii) an new algorithm for mutual information-based feature selection in supervised learning setting iii) an information-theoretic framework for learning disentangled and interpretable representations in unsupervised setting using deep neural networks. For the latter two tasks, we propose to use a variational lower bound for efficient estimation and optimization of mutual information. And for the last task, we have also made a substantial connection of the learning objective with variational auto-encoders (VAE).
Location: Social Sciences Building (SOS) - B37
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Kuan Liu
Mon, May 07, 2018 @ 01:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Candidate: Kuan Liu
Title: Scalable machine learning algorithms for implicit feedback based recommendation
Committee: Prem Natarajan, Kevin Knight, Shri Narayanan (outside member)
SAL 322
May 7
1-3pm
Abstract:
Whether in e-commerce, social networks, online music and TV, and many other modern online services, item recommendation stands out to be one of the most important algorithmic components. It recommends items to users that are useful and relevant. It makes huge economic values and is an important information filtering tool.
The primary goal of this thesis research is to provide machine learning solutions to item recommendation in large scale. The ever-increasing data volume and rich data formats have created a big gap between the requirements of modern recommender systems and our algorithm ability to handle large scale tasks. We work towards efficient personalized ranking algorithms to handle large data volume and advance content-based approaches to incorporate rich side information.
The thesis work mainly focuses on the following aspects towards this goal: (1) Novel ranking algorithms to deal with large itemsets (2) Deep learning methods to model sequential properties of user feedback (3) To incorporate heterogeneous attributes (4) To fuse signals from multiple modalities. In this talk, I will provide a brief overview of item recommendation history and our contributions. I will discuss our recent work on batch-based ranking algorithms for recommendation from large itemsets and our new methods to fuse signals from multiple modalities.
Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Xing Shi
Tue, May 08, 2018 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Candidate: Xing Shi
Date: May 8, 10am at SAL 322
Committee: Kevin Knight (chair), Jonathan May and Shri Narayanan
Abstract:
Recurrent neural networks (RNN) have been successfully applied to various Natural Language Processing tasks, including language modeling, machine translation, text generation, etc. However, several obstacles still stand in the way: First, due to the RNN's distributional nature, few interpretations of its internal mechanism are obtained, and it remains a black box. Second, because of the large vocabulary sets involved, the text generation is very time-consuming. Third, there is no flexible way to constrain the generation of the sequence model with external knowledge. Last, huge training data must be collected to guarantee the performance of these neural models, whereas annotated data such as parallel data used in machine translation are expensive to obtain. This work aims to address the four challenges mentioned above.
To further understand the internal mechanism of the RNN, we choose neural machine translation (NMT) systems as a testbed. We first investigate how NMT outputs target strings of appropriate lengths, locating a collection of hidden units that learns to explicitly implement this functionality. Then we investigate whether NMT systems learn source language syntax as a by-product of training on string pairs. We find that both local and global syntactic information about source sentences is captured by the encoder. Different types of syntax are stored in different layers, with different concentration degrees.
To speed up text generation, we propose two novel GPU-based algorithms: 1) Utilize the source/target words alignment information to shrink the target side run-time vocabulary; 2) Apply locality sensitive hashing to find nearest word embeddings. Both methods lead to a 2-3x speedup on four translation tasks without hurting machine translation accuracy as measured by BLEU. Furthermore, we integrate a finite state acceptor into the neural sequence model during generation, providing a flexible way to constrain the output, and we successfully apply this to poem generation, in order to control the meter and rhyme.
To improve NMT performance on low-resource language pairs, we re-examine multiple technologies that are used in high resource language NMT and other NLP tasks, explore their variations and result in a strong NMT system for low resource languages. Experiments on Uygher-English shows a 10+ BLEU score improvement over the vanilla NMT system.
Location: Henry Salvatori Computer Science Center (SAL) - 322
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense- Soravit (Beer) Changpinyo
Tue, May 08, 2018 @ 12:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Candidate: Soravit (Beer) Changpinyo
Committee: Fei Sha (chair), Kevin Knight, C.-C. Jay Kuo (outside member)
Title: Modeling, Learning, and Leveraging Similarity
Time & Place: Tuesday, May 8th, 12-2pm, SAL 213
Abstract:
Measuring similarity between any two entities is an essential component in most machine learning tasks. In this defense, I will describe my research work that provides a set of techniques revolving around the notion of similarity.
The first part involves "modeling and learning" similarity. We introduce Similarity Component Analysis (SCA), a Bayesian network for modeling instance-level similarity that does not observe the triangle inequality. Such a modeling choice avoids the transitivity bias in most existing similarity models, making SCA intuitively more aligned with the human perception of similarity.
The second part involves "learning and leveraging" similarity for effective learning with limited data, with applications in computer vision and natural language processing. We first leverage incomplete and noisy similarity graphs in different modalities to aid the learning of object recognition models. In particular, we propose two novel zero-shot learning algorithms that utilize class-level semantic similarities as a building block, establishing state-of-the-art performance on the large-scale benchmark with more than 20,000 categories. As for natural language processing, we employ multi-task learning (MTL) to leverage unknown similarities between sequence tagging tasks. This study leads to insights regarding the benefit of going to beyond pairwise MTL, task selection strategies, as well as the nature of the relationships between those tasks.
Location: Henry Salvatori Computer Science Center (SAL) - 213
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Sayan Ghosh
Tue, May 08, 2018 @ 02:00 PM - 04:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title : Multimodal Representation Learning of Affective Behavior
PhD Candidate: Sayan Ghosh
Date : 8th May , 2 PM PST
Venue: PHE 223
Committee : Prof. Stefan Scherer (Chair), Prof. Louis-Philippe Morency, Prof. Kevin Knight, Prof. Panayiotis Georgiou (EE)
Abstract:
With the ever increasing abundance of multimedia data available on the Internet and crowd-sourced datasets/repositories, there has been a renewed interest in machine learning approaches for solving real-life perception problems. However, such techniques have only recently made inroads into research problems relevant to the study of human emotion and behavior understanding. The primary research challenges addressed in this defense talk pertain to unimodal and multimodal representation learning, and the fusion of emotional and non-verbal cues for language modeling . There are three primary contributions of this dissertation -
(1) Unimodal Representation Learning: In the visual modality a novel multi-label CNN (Convolutional Neural Network) is proposed for learning AU (Action Unit) occurrences in facial images. The multi-label CNN learns a joint representation for AU occurrences, obtaining competitive detection results; and is also robust across different datasets. For the acoustic modality, denoising autoencoders and RNNs (Recurrent Neural Networks) are trained on temporal frames from speech spectrograms, and it is observed that representation learning from the glottal flow signal (the component of the speech signal with vocal tract influence removed) can be applied to speech emotion recognition.
(2) Multimodal Representation Learning: An importance-based multimodal autoencoder (IMA) model is introduced which can learn joint multimodal representations as well as importance weights for each modality. The IMA model achieves performance improvement relative to baseline approaches for the tasks of digit recognition and emotion understanding from spoken utterances.
(3) Non-verbal and Affective Language Models: This dissertation studies deep multimodal fusion in the context of neural language modeling by introducing two novel approaches - Affect-LM and Speech-LM. These models obtain perplexity reductions over a baseline language model by integrating verbal affective and non-verbal acoustic cues with the linguistic context for predicting the next word. Affect-LM also generates text in different emotions at various levels of intensity. The generated sentences are emotionally expressive while maintaining grammatical correctness as evaluated through a crowd-sourced perception study.
Location: Charles Lee Powell Hall (PHE) - 223
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense- Haifeng Xu
Wed, May 09, 2018 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Information as A Double-Edged Sword in Strategic Interactions
PhD Candidate: Haifeng Xu
Committee:
Shaddin Dughmi (Chair), Milind Tambe (Chair), David Kempe, Detlof von Winterfeldt, Vincent Conitzer, Odilon Camara.
Location & Time: SSL 150, 10 - 12 pm May 9th.
Abstract:
Strategic interactions among self-interested agents (a.k.a., games) are ubiquitous, ranging from economic activity in daily life and the Internet to defender-adversary interactions in national security. A key variable influencing agents' strategic decision making is the information they have available about their environment as well as the preferences and actions of others. In this talk, I will describe my work on computational questions pertaining to the role of information in games. In particular, I will illustrate the double-edged role of information through two threads of my research: (1) how to utilize information to one's own advantage in strategic interactions; (2) how to mitigate losses resulting from information leakage to an adversary. In each part, I will demonstrate how the study of fundamental theoretical questions sheds light on executable solutions to real-world problems in security applications including, e.g., delivered software to the Federal Air Marshal Service for improving the scheduling of US federal air marshals.
Location: 150
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense- Sean Mason
Wed, May 09, 2018 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Optimization Based Whole-Body Control and Reactive Planning for a Torque Controlled Humanoid Robot
PhD Candidate: Sean Mason
Date and Time: Wednesday May 9, 2018 at 10:00 AM in RTH 406
Committee: Stefan Schaal (Chair), Gaurav Sukatme, James Finley, Ludovic Righetti
Abstract:
Humanoid robots are expected to both locomote and interact with objects within unstructured environments. As robot hardware technologies have advanced, high-bandwidth, torque-controlled robots have become more widely-available as research platforms. In this work, I explore optimization-based methods for planning and control of a humanoid robot. I focus on the importance of controlling contact interactions with the environment for the tasks of balancing and walking of a bipedal system. This work is driven by and centered on the challenge of real robot implementations, and thus addresses the questions that come along with designing control algorithms for real systems. I will present lightweight control algorithms for whole-body balance, a model-predictive control approach to walking while using hands, and show extensive experiments on a humanoid robot.
Location: Ronald Tutor Hall of Engineering (RTH) - 406
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Kyriakos Zarifis
Wed, May 09, 2018 @ 01:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Making Web Transfers More Efficient
PhD Candidate: Kyriakos Zarifis
Date: 05-09-18
1pm
SAL 322
Committee:
Ethan Katz-Bassett (Chair)
Ramesh Govindan
Konstantinos Psounis (Outside)
Abstract:
Delays in web applications have been repeatedly shown to negatively impact business revenues. In this dissertation we perform studies related to Web transfer delays specific to propagation delay due to inflated paths, and delays in transferring data between servers and clients due to inefficient use of the communication channels.
Previous research has shown that the shortest path between a client and a server is not always selected, due to routing protocol policy-based decisions. We develop a methodology identify root causes of path inflation, specifically focusing on mobile traffic directed to Google servers, in order to understand the evolution of the infrastructure of mobile carrier networks and how it can affect user experience.
Once a connection has been established, information is exchanged between the two hosts according to rules defined by HTTP, the application layer protocol used for today's Web transfers. In this work we develop a model of the new version of HTTP/2 and pass through it a large dataset of HTTP/1 traces, in order to understand the performance implications of deploying the new version of the protocol in the wild. Our study exposes several opportunities for improvements, specifically using a new feature that allows a server to send to the client an object without the client requesting it. Generalizing from that observation, we design, develop and evaluate a system that allows CDNs to utilize idle network time around page downloads to send to the client content that the client is expected to request in the current or next page navigation. We show that if implemented correctly, speculative content prepositioning on the client can achieve a performance improvement comparable to having a page loaded on the client cache.Location: 322
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Aaron Schlenker
Mon, May 14, 2018 @ 10:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
University Calendar
Title: Game Theoretic Deception and Threat Screening for Cyber Security
PhD Candidate: Aaron Schlenker
May 14th
10am
SSL 150
Committee:
Milind Tambe (Chair)
Jelena Mirkovic
Jonathan Gratch
Muhammed Naveed
Richard John
Abstract:
Protecting an organization's cyber assets from intrusions and breaches due to attacks by malicious actors is an increasingly challenging and complex problem. Companies and organizations who operate enterprise networks deploy various software and tools to protect from these attacks, such as anti-virus software and Intrusion and Detection Systems (IDS), along with dedicated teams of cyber analysts tasked with the general protection of an organization's cyber assets. In order to compromise a network, an adversary must complete the Cyber Kill Chain which is a series of steps outlining the components of a successful cyber breach. During the Cyber Kill Chain, there are numerous opportunities for the network administrator (defender) to intercept the adversary and thwart an attack. In this talk, I will describe how computational game theory can be used to capture the interaction between the adversary and network administrator in cyber security along with two potential applications of game theory to problems faced by the network administrator to optimize the use of their limited security resources. The first application proposes a framework for deceiving cyber adversaries during the reconnaissance phase of an attack and I will describe a model that provides deceptive strategies to the defender that lead to hackers attacking non-critical systems in the defender's network. The Second application corresponds to the prioritization of alerts generated from Intrusion Detection and Prevention systems throughout a network and I will describe a model that accounts for various salient features in cybersecurity alert allocation when determining the best strategies for the network administrator.
Location: Seaver Science Library (SSL) - 150
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
PhD Defense - Arman Shahbazian
Thu, May 17, 2018 @ 11:00 AM - 01:00 PM
Thomas Lord Department of Computer Science
University Calendar
PhD Candidate: Arman Shahbazian
Committee: Nenad Medvidovic (chair), Chao Wang, Sandeep Gupta (outside member)
Title: Techniques for Methodically Exploring Software Development Alternatives
Date and Time: Thursday, May 17th, 11:00am-1:00pm
Location: VKC 210
Abstract:
Designing and maintaining a software system's architecture typically involve making numerous design decisions, each potentially affecting the system's functional and nonfunctional properties. Understanding these design decisions can help inform future decisions and implementation choices, and can avoid introducing architectural inefficiencies later. Despite their importance, the support for engineers to make these decisions is generally lacking. There is a relative shortage of techniques, tools, and empirical studies pertaining to architectural design decisions. Moreover, design decisions are rarely well documented and are typically a lost artifact of the architecture creation and maintenance process. The loss of this information can thus hurt development. To address these shortcomings, we develop a set of techniques to enable methodical exploration of such decisions and their effects. We develop a technique, named RecovAr, for automatically recovering design decisions from the project's readily available history artifacts, such as an issue tracker and version control repository. Building on RecovAr, we create PredictAr that aims to prevent the consequences of inadvertent architectural change. The result of such changes is accumulation of technical debt and deterioration of software quality. In this dissertation we take a step toward addressing that scarcity by using the information in the issue and code repositories of open-source software systems to investigate the cause and frequency of such architectural design decisions. We develop a predictive model that is able to identify the architectural significance of newly submitted issues, thereby helping engineers to prevent the adverse effects of architectural decay. We close the loop by helping engineers to not only predict and recover architectural design decisions, but also make new design decisions that are informed and well-considered. To that end, we present eQual, a novel model-driven technique for simulation-based assessment of architectural designs that helps architects understand and explore the effects of their decisions.
Location: 210
Audiences: Everyone Is Invited
Contact: Lizsl De Leon
-
The Connected Hospital: Preparing for 21st Century Healthcare and Community Engagement
Mon, May 21, 2018 @ 11:00 AM - 12:00 PM
Alfred E. Mann Department of Biomedical Engineering
University Calendar
Today-�s fundraising, marketing and patient engagement environment for healthcare institutions is more complex than ever. Prospective donors, grateful patients and their families, new patients, and health and wellness advocates demand a personalized, omni-channel relationship. How can we give that to them cost affordably and efficiently? We want to help!
Join healthcare data, innovation & technology, and marketing & branding experts, Michael Johnston from hjc, Russ Cobb from Blackbaud, and Dr.George Tolomiczenko, PhD from USC for a half day workshop to advance your team's shift to a more digitally connected, omni-channel and supporter/patient-centric future.
You'll hear from Dr. George Tolomiczenko, Administrative Director of The Health, Technology & Engineering Program (HTE@USC) at University of Southern California. He-�ll outline how technology and innovation will play a key role for hospitals and their foundations now and in the future. And to emphasize his point, he-�s invited a cutting-edge health technology start up Stasis Labs to share how their hardware and software solution is poised to impact healthcare here and around the world. We think it-�s a wonderful, intimate look inside innovation in our sector.
Contact: Nadine Afari
nafari@usc.edu to receive RSVP link
Location: The USC Norris Comprehensive Cancer Center, Room: NRT LG 503
Audiences: Everyone Is Invited
Contact: Nadine Afari
