Back to the workshop page

Adam Baker: Systems Research in Huawei

Abstract

This talk will give an overview of the newly established Systems Infrastructure Research (SIR) lab, part of the Edinburgh Research Centre. I will present our core research areas and describe current and future projects in distributed systems and operating systems.

Bio

Adam Barker is the Director of the Systems Infrastructure Research (SIR) lab in the Huawei Edinburgh Research Centre and a Professor of Computer Systems at the School of Computer Science, University of St Andrews. Adam’s core research focuses on distributed systems, he has published over 80 peer-reviewed papers and his research is supported by over £3 Million in grant income from a variety of sources including the EPSRC and Royal Society. Adam has spent large portions of his career in industry and has worked as a Research Scientist at Google in the Bay Area on two separate occasions where he contributed to Borg and Google Cloud Platform (GCP).

Back to the workshop page

Dawud Hage: QA in petal search: challenges, opportunities and future directions

Abstract

Modern search engines need to better understand what users are looking for rather than being a simple index of webpages. For the information seeking need, question answering plays an important role in fullfilling this need. In this talk I would like to present QA systems that we build as well as the challenges and future directions that we want to explore to further improve the need of our users.

Bio

Dawud leads the Language Understanding team at the Huawei research center in Amsterdam. The team focusses on question answering, information extraction and recently conversational systems. Prior to Huawei he worked at trivago on multiple machine learning projects from search to recommender systems.

Back to the workshop page

Glynn Winskel:  Making Concurrency Functional

Abstract

Bio

Back to the workshop page

Jeff Pan: Knowledge Graphs and Search: Past, Present and Future

Abstract

Knowledge Graphs can help search engines  to find the right things, get the best summary and many more. In this talk, I will share some thoughts on the past, present and future of knowledge graphs and search, including some related challenges that we might look into in the Knowledge Graphs Lab at the Edinburgh Research Centre.

Bio

Jeff Pan is the Director of the Knowledge Graphs Lab at the Huawei Edinburgh Research Centre. He is a Reader in the School of Informatics at the University of Edinburgh. He helps chair the Knowledge Graphs Group at the Alan Turing Institute. He is a receiver of the Changjiang Scholar Award (2019).

Back to the workshop page

Maciej Zembrzuski: Petal Maps: challenges, opportunities and future directions

Abstract

Natural language processing and understanding is a crucial part of the Huawei Petal Map Search with a wide impact on the product from language identification to spelling error correction. The quality expectations are high because users want nothing but a product that is better compared to our top competitors.  This imposes great challenges for us in developing solutions that must demonstrate both exceptional performance and execution speed.

Bio

Maciej Zembrzuski leads Mas Search team in Huawei R&D center in Amsterdam. He created several NLP solutions which have been productionized by Huawei and in the past by other industry leaders. His interest covers wide range of NLP tasks with particular focus on end-to-end multilingual solutions. He takes pride in combining development with research.

Mark Steedman: Semantic Parsing for Conversational Question-Answering

Abstract

Bio

Back to the workshop page

Roberto Ierusalimschy:  ​ Lua and Pallene  

Abstract

Bio

Back to the workshop page

Shubhabrata Roy: Challenges and research opportunities of Natural Scene Text Recognition: A use case of automatic POI name extractions for Petal Maps

Abstract

Points Of Intersts (POI) are key features of any modern map platform. They improve user experience in terms of navigation, place search etc. In Huawei we currently acquire POI data from different providers, but we also want to build capability to extract POI information from images and videos. In this talk we would present the challenges like text attribute decision, occlusion etc. of Natural Scene Text Recognition for extracting POI information in the wild.

Bio

Back to the workshop page

Talaat Khalil: MT in petal search: challenges and opportunities

Abstract

Translation technology is an integral part in Huawei search environment. Direct applications are diverse and range from open domain input like translation box and web translation to domain specific applications like the application to the shopping vertical. Moreover, the input itself can be dependent on other systems like OCR or ASR. This poses a lot of challenges regarding accuracy, robustness, language scalability and serving performance which we would like to present about.

Bio

Back to the workshop page

Yang Cao: Querying Distributed and Shared Data

Abstract

The need for querying shared data has been increasingly evident due to the many data sharing initiatives from e-government, healthcare, finance and the AI industry, among other things. Shared data typically comes in the form of distributed datasets. Different from conventional homogeneous distributed databases, shared data however is heterogeneous in terms of data models and communication patterns. In this talk, we will discuss approaches to querying shared data with both types of heterogeneities. We will cover a model that captures heterogeneous communication restrictions and costs in query processing, a method for linking relations and graphs, and an extension to SQL with the added capacity of querying relational and graph datasets within the same system or even in a single query. If time allows, we will also briefly cover the database research at Edinburgh in a broad sense that is relevant to the Jointlab and Huawei.

Bio

Yang Cao is a Chancellor’s Fellow at the University of Edinburgh. He has been working on database query processing, graph querying, and systems and theory of data management in general. He holds a number of awards and fellowships, including the RAEng research fellowship (2020), SIGMOD Research Highlight award (2018), and SIGMOD best paper award (2017).

Back to the workshop page

Mike O’Boyle: Rethinking optimizing compiler technology

Abstract

Moore’s Law has been the main driver behind the extraordinary success of computer systems. However, with the technology roadmap showing a decline in transistor scaling, computer systems are increasingly specialised and diverse. As it stands, software will simply not fit and current compiler technology is incapable of bridging the gap. We need to fundamentally rethink the role of the compiler. This talk describes some novel approaches to automatic optimization of legacy softwarewith minimal user involvement. It also provide an overview of ongoing work at Edinburgh in compiler technology and computer architecture.

Bio

Back to the workshop page

Jinyun (Joey) Ye:  Challenges in Huawei Compiler

Abstract

Huawei pushes the boundary of how compiler can better serve the purpose of improving productivity, security and performance of programming. In this front we are facing many technical challenges from frontend, codegen, optimization to debugging, . In this talk I will introduce some of these challenges and seek help from academia to break through them.

Bio

Back to the workshop page

Philip Wadler:  GATE: Gradual Algebraic Effect Types

Abstract

Two recent exciting trends in programming languages are gradual types and algebraic effect handlers.  Several steps are required to bring algebraic effect handlers to wider use, one of the most important being the development of a suitable gradual type system.

Bio

Back to the workshop page

Nikos Ntarmos: (Graph) Database Challenges @ Huawei ERC: Past, Present and Future

Abstract

This talk will provide a high-level overview of the research problems tackled by the Database Lab at Huawei’s Edinburgh RC over the past couple of years, as well as an outlook of what lies ahead. In doing so, it will also attempt to list areas of interest and, hopefully, solicit further discussions for possible future collaborations.

Bio

Dr Nikos Ntarmos is the Director of Database Lab at the Edinburgh Research Centre and a Senior Lecturer at the School of Computing Science, U. of Glasgow. His research interests lie in the area of large-scale distributed data management systems, with current research work focusing on issues pertaining to storage, indexing and query processing in distributed data stores, (graph) database systems, geo-distributed data management infrastructures, and joint in-rest and streaming data processing.

Back to the workshop page

Dan Ghica: Terms, diagrams, graphs — a syntactic trinity

Abstract

Syntax is a data structure that comes in many shapes and forms. Humans prefer its serialized representation, which we call “text”, but this form is not ideal for reasoning about its properties, nor for processing it algorithmically. Compilers have long used graph-like data structures to represent syntax (ASTs) and related concepts, especially data flow and control flow graphs, but the connection with the textual form was not made systematic enough. In this talk I will introduce the third and lesser known member of the syntactic trinity, “string diagrams”, and show how they serve as a bridge between the textual and graph representations. Moreover, string diagrams help us derive an improved graph-like representation for syntax which we call “hierarchical hypergraphs” (or “hypernets”), with applications to analysis, optimisation, and transformation of intermediate code in compilers.

Bio

Dan R. Ghica is the Director of the Programming Languages Laboratory in the Huawei Edinburgh Research Centre, since April 2020. He is also a Professor of Semantics of Programming Languages at the University of Birmingham. His principal area of research is semantic-directed approaches to programming language verification and compilation, in particular using game semantics and sub-structural type systems, with applications to high level synthesis (making circuits out of functional programs) and heterogeneous compilation.

Back to the workshop page

Titles and Abstracts of the Talks at the Huawei Edinburgh Workshop in December 2021 / Huawei-Edinburgh Joint Lab by blogadmin is licensed under a Creative Commons Attribution CC BY 3.0