Monday, 5/8/2019
09:00 – 09:20 Welcome coffee
09:20 – 09:30 Opening remarks
Session Chair: Ashraf Abdul
09:30 – 10:15 Automated Program Repair - Abhik Roychoudhury
10:15 – 10:40 Group Fairness and Diversity in the Allocation of Indivisible Goods - Mithun Chakraborty
10:40 – 11:05 Towards Robust ResNet - Jingfeng Zhang
11:05 – 11:15 Break
11:15 – 11:40 IMLI: An Incremental Framework for MaxSAT-Based Learning of Interpretable Classification Rules - Bishwamittra Ghosh
11:40 – 12:05 Designing Theory-Driven User-Centric Explainable AI - Danding Wang
12.05 – 13:00 Lunch Break
Session Chair: Muoi Tran
13:00 – 13:45 Training-by-Fitting: Self-Supervision for 3D Hand Pose Estimation - Angela Yao
13:45 – 14:10 Learning to Learn from Noisy Labeled Data - Junnan Li
14:10 – 14:35 One Engine To Serve'em All: Inferring Taint Rules Without Architectural Semantics - Zheng Leong Chua
14:35 – 15:00 Coffee break
15:00 – 15:25 Neuro-Symbolic Execution: Augmenting Symbolic Execution with Neural Constraints - Shiqi Shen
15:25 – 15:50 A Semantic Touch Interface for Flying Camera Photography - Lan Ziquan
15:50 – 16:15 Enhancing Stock Movement Prediction with Adversarial Training - Fuli Feng
16:15 – 16:40 Multi agent system verification - Dileepa Fernando
16:40 – 17:00 Break
17:00 – 18:15 Panel 1 - Industry/Academics after PhD: prospects and challenges (Moderator : Aashish Kolluri)
Panelists:
– Prateek Saxena
– Kuldeep S. Meel
– David S. Rosenblum
– Jia Yaoqi, Chief Technology Officer & Co-founder at Zilliqa
– Jagannadan Varadarajan, Head of Data Science (Machine Learning) at Grab
18:15 – 19:00 Dinner
Tuesday, 6/8/2019
09:00 – 09:30 Welcome coffee
Session Chair: Himeshi De Silva
09:30 – 10:15 Adversarial Machine Learning: Challenges of building trust in AI - Reza Shokri
10:15 – 10:40 Better Care Quality at Lower Cost: Should I Participate in the Online Healthcare Communities - Luo Kai
10:40 – 11:05 Nudge Consumers through the Manipulation of Price Expectation: Evidence from a Field Experiment - Gao Yuting
11:05 – 11:15 Break
11:15 – 11:40 Facilitating mental product interaction: the effects of product orientation and computer interface - Peng Xixian
11:40 – 12:05 The Impact of Peer Influence on Academic Performance: A Three-Stage Co-Evolution Model - Ding Dan
12:05 – 13:00 Lunch Break
Session Chair: Mithun Chakraborty
13:00 – 13:45 Online Investment Advice and Micro-Celebrities: An Empirical Analysis of a Social Investment Platform - Goh Khim Yong
13:45 – 14:10 Want to Play DASH? A Game Theoretic Approach for Adaptive Streaming over HTTP - Abdelhak Bentaleb
14:10 – 14:35 A Stealthier Partitioning Attack against Bitcoin Peer-to-Peer Network - Muoi Tran
14:35 – 15:30 Poster Session + Break
15:30 – 15:55 Exploiting the Laws of Order in Smart Contracts - Aashish Kolluri
15:55 – 16:20 Towards Scaling Blockchain Systems via Sharding - Hung Dang
16:20 – 16:45 Introduction to Non-Malleable Codes and Inception Coding - Maciej Obremski
16:45 – 17:30 Break
17:30 – 18:30 Orientation for new graduate students
18:30 – 19:00 Research award presentation
19:00 – 19:30 Program briefing for PhD Students and MComp students
19:30 – 20:30 Dinner
Wednesday, 7/8/2019
09:00 – 09:30 Welcome coffee
Session Chair: Tapas Nayak
09:30 – 10:15 An Optimal Algorithm for Heavy Hitters in Insertion Streams and Related Problems - Arnab Bhattacharyya
10:15 – 10:40 PR3: Power Efficient and Low Latency Baseband Processing for LTE Femtocells - Nishant Budhdev
10:40 – 11:05 SoundUAV: Fingerprinting Acoustic Emanations for Delivery Drone Authentication - Soundarya Ramesh
11:05 – 11:15 Break
11:15 – 11:40 Friot: A Functional Reactive Language for IoT Programs with Dependent Type-and-Effect System - Yahui Song
11:40 – 12:05 Precise Time-synchronization in the Data-Plane using Programmable Switching ASIC - Pravein Govindan Kannan
12:05 – 13:00 Lunch Break
Session Chair: Sangharatna Godboley
13:00 – 13:45 Emptiness doesn't have to mean nothingness; haziness doesn’t have to mean iffiness - Lim Soon Wong
13:45 – 14:10 Crash-avoiding Program Repair - Gao Xiang
14:10 – 14:35 HyCUBE : A CGRA with Reconfigurable Single-cycle Multi-hop Interconnect - Manupa Karunaratne
14:35 – 15:00 Coffee break
15:00 – 15:25 Learning parametric skills for autonomous driving - Gao Wei
15:25 – 15:50 GPU-based Graph Traversal on Compressed Graphs - Sha Mo
15:50 – 16:15 Topological Data Analysis with ε-net Induced Lazy Witness Complex - Naheed Anjum Arafat
16:15 – 16:40 Data-driven methods for knowledge discovery in Regulomics - Stefano Perna
16:40 – 17:00 Break
17:00 – 18:15 Panel 2 - Different aspects of building a smart city (Moderator : Alan Tsang)
Panelists:
– Tulika Mitra
– Jun Han
– Yair Zick
– David Hsu
– Beng Chin Ooi
18:15 – 19:00 Dinner
Thursday, 8/8/2019
09:00 – 09:30 Welcome coffee
Session Chair: Pravein Govindan Kannan
09:30 – 10:15 How to Video Stream on the Internet like the Pros - Roger Zimmermman
10:15 – 10:40 Knowledge-aware Multimodal Dialogue Systems - Lizi Liao
10:40 – 11:05 Learning to Detect Human-Object Interactions with Knowledge - Bingjie Xu
11:05 – 11:15 Break
11:15 – 11:40 Quantum Log-Approximate-Rank Conjecture is also False - Naresh Boddu
11:40 – 12:05 Weighted fair caching: Occupancy-centric allocation for space-shared resources - Lianjie Shi
12:05 – 13:00 Lunch Break
Session Chair: Vipul Arora
13:00 – 13:45 Deductive Synthesis of Heap-Manipulating Programs: Sound, Expressive, Fast - Ilya Sergey
13:45 – 14:10 Sublinear Time Nearest Neighbor Search over Generalized Weighted Space - Lei Yifan
14:10 – 14:35 Sequicity: Simplifying Task-oriented Dialogue Systems with Single Sequence-to-Sequence Architectures - Wenqiang Lei
14:35 – 15:30 Poster Session + Break
15:30 – 15:55 Scaling Data Stream Processing on Shared-Memory Multicore Architectures - Shuhao Zhang
15:55 – 16:20 Monitoring the Evolving Congestion Control Landscape of the Internet - Ayush Mishra
16:20 – 16:45 Improving Join Reorderability with Compensation Operators - Wang Taining
16:45 – 17:00 Music and Wearable Computing for Health and Learning - Wang Ye
17:00 – 17:05 Closing remarks
17:05 – 19:00 Dinner
19:00 – 20:00 Concert by Sound and Music Computing Lab
Monday, 5/8/2019, 09:30 – 10:15
Automated Program Repair – Abhik Roychoudhury
Automated program repair is an emerging and exciting field of research, which allows for automated rectification of errors and vulnerabilities. The use of automated program repair can be myriad, such as (a) improving programmer productivity (b) automated fixing of security vulnerabilities as they are detected, (c) self-healing software for autonomous devices such as drones, as well as (d) use of repair in introductory programming education by grading and providing hints for programming assignments. One of the key technical challenges in achieving automated program repair, is the lack of formal specifications of intended program behavior. In this talk, we will conceptualize the use of symbolic execution approaches and tools for extracting such specifications. This is done by analyzing a buggy program against selected tests, or against reference implementations. Such specification inference capability can be combined with program synthesis techniques to automatically repair programs. The capability of specification inference also serves a novel use of symbolic execution beyond verification and navigation of large search spaces. Automated program repair via symbolic execution goes beyond search-based approaches which attempt to lift patches from elsewhere in the program. Such an approach can construct “imaginative” patches, serves as a test-bed for the grand- challenge of automated programming, and contributes to the vision of trustworthy self-healing software.
Monday, 5/8/2019, 13:00 – 13:45
Training-by-Fitting: Self-Supervision for 3D Hand Pose Estimation - Angela Yao
Learning based hand pose estimation methods, especially deep learning based
methods, requires large amount of accurate annotations on real-world data to
achieve high accuracy. However, acquiring such accurate annotated samples
can be extremely difficult and expensive. To mitigate the dependency on
large amounts of annotation, we propose to leverage unlabeled samples
instead. We propose a method that bridges model-based optimization and
discriminative learning by using model-fitting errors to train deep neural
networks. We demonstrate the ability of our method to generalize on two
different models, one from a set of spheres and one from triangular meshes.
Our proposed method makes highly accurate pose estimates comparable to
current supervised methods and advances state-of-the-art in unsupervised
learning for hand pose estimation.
Tuesday, 6/8/2019, 09:30 – 10:15
Adversarial Machine Learning: Challenges of building trust in AI - Reza Shokri
Machine learning algorithms have shown an unprecedented predictive power for many complex learning tasks. As they are increasingly being deployed in large scale critical applications for processing various types of data, new questions related to their trustworthiness would arise. Can machine learning algorithms be trusted to have access to individuals' sensitive data? Can they be robust against noisy or adversarially perturbed data? Can we reliably interpret their learning process, and explain their predictions? In this talk, I will go over the challenges of building trustworthy machine learning algorithms in centralized and distributed (federated) settings, and will briefly discuss the inter-relation between privacy, robustness, and interpretability.
Tuesday, 6/8/2019, 13:00 – 13:45
Online Investment Advice and Micro-Celebrities: An Empirical Analysis of a Social Investment Platform - Goh Khim Yong
Retail investors are increasingly seeking online investment advice from
social media. In this study, we conduct an empirical analysis of a social
investment platform (SIP), which relies on social media advisors (SMAs) to
contribute online investment commentaries and advice to retail investors. We
then simultaneously quantify the impact of both SMAs’ informational factors
and micro-celebrity tactics on their remuneration performance. This enables
us to derive the relative economic significance of these factors and infer
the information preferences of retail investors. Utilizing a dataset from a
Chinese SIP, our study performs an in-depth content analysis to measure the
informational factors of online investment advice and micro-celebrity
tactics of SMAs. We then propose a hierarchical Bayesian modelling framework
that accounts for various empirical issues: SMA heterogeneity,
self-selection of paid content generation, omitted variable bias and
measurement errors. Our findings demonstrate that SMAs achieve higher
remuneration performance in terms of subscription revenue if their
investment advice has more diverse sectors (diversity), more popular stocks
(popularity), fewer negative sentiments (negativity), higher short-term
predictive accuracy (accuracy), and more efforts in monitoring stocks across
periods (sustenance). Specifically, sentiment negativity has the strongest
relative effect. A one standard deviation increase in negativity leads to
RMB 43.29 change in the remuneration performance, followed by diversity
(2.81), sustenance (2.43), and popularity (1.84). In addition, an
affiliation-based micro-celebrity tactic is more effective than an
intimacy-based one. These two tactics also negatively moderate the effects
of informational factors such as diversity and intensity. Our findings
provide implications for academic research and practice.
Wednesday, 7/8/2019, 09:30 – 10:15
An Optimal Algorithm for Heavy Hitters in Insertion Streams and Related Problems - Arnab Bhattacharyya
We give the first optimal bounds for returning the heavy hitters in a data stream of insertions, together with their approximate frequencies, closing a long line of work on this problem. We show a randomized algorithm whose space complexity improves upon Misra and Gries' classic algorithm and is furthermore optimal upto constants. A modification of our algorithm can be used to estimate the maximum frequency up to an additive error. We also introduce several variants of the heavy hitters and maximum frequency problems, inspired by rank aggregation and voting schemes, and show how our techniques can be applied in such settings. Unlike the traditional heavy hitters problem, some of these variants look at comparisons between items rather than numerical values to determine the frequency of an item.
Joint work with Palash Dey (IIT Kharagpur) and David Woodruff (IBM Almaden)
Wednesday, 7/8/2019, 13:00 – 13:45
Emptiness doesn't have to mean nothingness; haziness doesn’t have to mean iffiness - Lim Soon Wong
Biological and clinical omics data are full of data holes, obfuscating batch effects, confounding dependencies, etc. These issues present interesting pervasive challenges in analyzing biological and clinical omics data. I will present some instances of these challenges, as well as our solutions to them. I will highlight the interplay of logic, statistics, and biological/physical principles in the development of these solutions.
Thursday, 8/8/2019, 09:30 – 10:15
How to Video Stream on the Internet like the Pros - Roger Zimmermman
Globally, video streaming currently accounts for about 70% of all
traffic on the internet and it is expected to reach 82% by 2021. Thus,
managing video traffic efficiently is of high importance for many
organizations. Furthermore, video communication is becoming an important
and integral part of many applications, for example, for staff training
or customer support calls. In recent years HTTP (Hypertext Transfer
Protocol) adaptive streaming (HAS) is being adopted with increasing
frequency and has become the de-facto standard for video streaming.
Thus, HAS technology has seen rapid and significant developments
recently. With the large-scale deployments of HAS, new and interesting
challenges have emerged. For example, the client-driven, on-off
adaptation behavior of HAS results in uneven bandwidth competition and a
decrease in end-user quality of experience (QoE). Also, the long latency
of HAS is problematic in live streaming. In this talk I will present
some of our proposals and recent work which aim to alleviate these
scalability issues.
Thursday, 8/8/2019, 13:00 – 13:45
Deductive Synthesis of Heap-Manipulating Programs: Sound, Expressive, Fast - Ilya Sergey
In my talk, I will describe a deductive approach for synthesising imperative programs with pointers from declarative specifications expressed in Separation Logic. The approach treats logical program specifications, given in the form or pre- and postconditions with a pure and a spatial part, as proof goals, and provides an algorithm for rule-directed program construction based on the shape of the symbolic heap footprint a desired program manipulates.
The program synthesis algorithm rests on the novel framework of Synthetic Separation Logic (SSL). The produced executable programs are correct by construction, in the sense that they satisfy the ascribed specifications, and are accompanied by complete proof derivations (i.e., certificates), which can be checked independently by a third-party verifier. The approach has been implemented as a proof search engine for SSL in a form a program synthesiser. For efficiency, the engine exploits properties of SSL specifications, aggressively relying on a version of the Frame Rule and commutativity of separating conjunction to prune the search space.
I will explain and showcase the use of SSL on characteristic examples, describe the design of out tool, and report on the experience of using it to synthesise a series of benchmark programs manipulating with heap-based linked data structures. I will also tell about some recent advances of using SSL-based synthesis, enhanced with immutability annotations, as well as its applications for program repair.
Thursday, 8/8/2019, 17:00 – 17:45
Music and Wearable Computing for Health and Learning - Wang Ye
Music is more than just a source of entertainment. Parents sing nursery rhymes to their young children to help them learn their first language; music therapists use music to help patients recover. I will present our recent research work in creating music computing technology specifically geared towards health and learning applications. In particular, I will introduce our Singing and Listening to Improve Our Natural Speaking (SLIONS) Karaoke smartphone app, which is designed to promote joyful learning of a second language. Wearable sensors are applied in many different domains. I will present our wearable sensor systems specially designed for quantifying patients’ gaits, as well as for facilitating physical exercise via dancing and music making. Towards this end, I will highlight our Mobility Analysis (MANA) smart-sensor system which is designed to quantify gaits of people with Parkinson’s disease, as well as Motion Initiated Music Ensemble with Sensors (MIMES) system which is designed to convert human movements to musical sounds for neurologic music therapy.
