When: Monday, April 20, 2020
Where: Online Presentation - Google Meet: https://meet.google.com/qrc-ieyu-bhs
Time: 11:00 AM

Focus on Research (FoR) is an opportunity for any COSC Ph.D. student to discuss a research project (with or without preliminary results), a conference dry run, or any research topic of interest to present to an audience of peers and faculty. It is a great avenue for Ph.D. students to practice presentation skills in front of a larger and broader audience.

Application Agnostic Network Traffic Modeling for Realistic Traffic Generation

Oluwamayowa Ade Adeleke, Ph.D. Student

Abstract:

Research and testing in networking often require experimentation with real or representative network traffic. Privacy policies often limit access to production traffic by third parties, which includes most academic researchers. Presently, most experimenters rely on synthetic traffic generators that send packets at fixed rates, or at rates based on statistical distributions. Others replay captured packets' traces which often have limited durations. In response to this, we propose to create ‘traffic models’ for the patterns of network traffic seen in production networks by using a combination, clustering algorithms in conjunction with stochastic and empirical distributions, to model applications’ network behaviors from traffic traces, after removing all protocol-specific reactions to network impairments. In this talk, we discuss our current progress, describing our background survey of traffic generators, and how our current implementation of the system creates traffic models for realistic traffic generation in test beds. The outcome and methods derived from this research promises to improve the how experimentation on large scale networks (datacenter, cloud, enterprise, and IOT) is done, especially in academia.

Bio:

Oluwamayowa Ade Adeleke is a 4th-year Ph.D. student, advised by Dr. Deniz Gurkan. His interests lie in computer networks, cloud computing, traffic modeling and applications of machine learning to computer networks.

Learning to learn: Meta-Learning in Machine Learning

Mikhail Mekhedkin-Meskhi, Ph.D. Student

Abstract:

Learning new models in machine learning for each problem is an expensive process. It requires expert knowledge, domain knowledge, data pre-processing and model building. This can be a lengthy process overall. Meta-learning is the process of exploiting past experience on problems to solve a newly unseen problem without starting from scratch. In this talk, I will focus on presenting the meta-learning paradigm, current approaches, and what lies ahead for the field.

Mikhail Mekhedkin-Meskhi is a first year Ph.D. student at the Pattern Analysis Lab under the supervision of Dr. Ricardo Vilalta. Mekhedkin-Meskhi's research focus is on meta-learning, transfer learning and statistical learning theories.

Attending the Emotions to Detect Online Abusive Language

Niloofar Safi Samghabadi, Ph.D. Student

Abstract:

In recent years, abusive behavior has become a serious issue in online social networks. In this talk, we present a new corpus for the task of abusive language detection that is collected from a semi-anonymous online platform and, unlike most of the other available resources, is not created based on profanities. We develop a benchmark system to incorporate emotions into textual cues to improve aggression identification. We evaluate our proposed method on different corpora and show new state-of-the-art results with respect to abusive language detection.

Bio:

Niloofar Safi Samghabadi is a Ph.D. student at the University of Houston. She works under the supervision of Dr. Thamar Solorio. Her research focuses on natural language processing, and she works on abusive language and cyberbullying detection.

Let Me Choose: From Verbal Context to Font Selection

Amirreza (Reza) Shirani, Ph.D. Student

Abstract:

In this study, we aim to learn associations between visual attributes of fonts and the verbal context of the texts they are typically applied to. Compared to related work leveraging the surrounding visual context, we choose to focus only on the input text, which can enable new applications for which the text is the only visual element in the document. We introduce a new dataset, containing examples of different topics in social media posts and ads, labeled through crowd-sourcing. Due to the subjective nature of the task, multiple fonts might be perceived as acceptable for an input text, which makes this problem challenging. To this end, we investigate different end-to-end models to learn label distributions on crowd-sourced data in order to capture intersubjectivity across all annotations.

Bio:

Amirreza (Reza) Shirani is a 4th-year Ph.D. student working on multiple natural language processing topics such as emphasis selection, font prediction, and community question-answering.