This Expert Meeting will take place at the ZHAW premises in Lagerstrasse 45, 8004 Zurich in room ZL O3.01 on the third floor (online participation is also possible for those who prefer this option) on Wednesday, May 10 from 17:30-19:00. After the meeting, there will be an apéro so that you can carry on your discussions and get to know each other.
We have the following two talks confirmed:
End-to-end ASR for Swiss German at Microsoft: A Transducer Approach Oscar Koller, Applied Scientist at Microsoft
Automatic speech recognition (ASR) for Swiss German is a challenging task due to the lack of a standardized writing system and the high regional variability of the dialects. In this talk, we present our work on developing end-to-end ASR models for Swiss German at Microsoft using transducer architectures. We show that transducers outperform hybrid models by over 20% in word error rate on a multi-dialectal corpus of Swiss German speech. We also compare our models to Whisper, a state-of-the-art sequence-to-sequence model for low-resource ASR, and find that transducer models achieve comparable results with much smaller model size and training time. Finally, we discuss how end-to-end models produce transliterations of Swiss German words instead of standard German translations affecting the readability and usability of the output and propose solutions to this problem.
Revolutionizing Natural Interaction with Swiss German: A Glimpse into the Future of Conversational AI Claudio Paonessa and Yanick Schraner, Researchers at FHNW
Get ready for a glimpse into the future of natural interaction with computer systems in Swiss German! We leveraged the latest advancements in speech-to-text and text-to-speech technology to create an engaging and interactive experience that showcases the results of our cutting-edge research.
Exploring the Acceptance of Intelligent Voice Assistants in Home Care Applications: Opportunities and Obstacles [10 mins presentation, 10 mins discussion] Edith Birrer, Researcher at iHomeLab – HSLU (Hochschule Luzern)
In the scope of co-creation sessions, care workers provided insights on applications and on concerns about Intelligent Voice Assistants (IVA) in the home of their clients or patients. The sessions focused on the potential to support the care documentation process by IVA. Participants’ expectations and worries spanned from the ability to handle dialects, to confidentiality issues, to integration in existing care documentation systems. However, there is a general openness toward the idea to employ IVA as means to improve the quality of care. The challenge foreseen for using IVA is to become as time efficient as care documentation systems in place. Alternatively, as suggested by participants, IVA could complement existing processes or even create new ones in the care context.
If you want to join, please fill in the following registration form by April 27: https://forms.gle/PmRQENtY8aybJeby5 Please note that the registration form includes information for the SwissNLP General Assembly which is co-located.
The Data Innovation Alliance’s second Expert Day in March 2023 was a hub of activity as experts from four key areas – Smart Maintenance, NLP & AI Technology, Spatial Data, and Smart Services – gathered to share their insights and mingle with researchers and industry professionals. The event kicked off with leaders from each Expert Group pre-discussing their plans for 2023, generating a wealth of innovative ideas for joint events and initiatives, and paving the way for exciting collaborations in the (near) future.
But that’s not all! The NLP and Digital Health groups are teaming up to bring you joint events that will revolutionize the way we approach data. And with the next Expert Day set for August 2023, featuring four expert groups once again, get ready for even more ground-breaking discussions and initiatives, organized jointly with other Innovation Boosters. Keep an eye on our events calendar for more information.
While the keynote speech may not have met expectations in terms of insights, it set the stage for what was to come – dynamic discussions and collaborations in the expert group break sessions. To ensure everyone had access to the wealth of information shared, short summaries of the discussions were written by participants in each room.
In short, the second Expert Day was a superb success, bringing together a diverse group of experts to debate their ideas and shape the future of data innovation.
Smart Services for Sustainability – Circular Servitization by Jürg Meierhofer
The Smart Services for Sustainability – Circular Servitization discussion was a dynamic conversation among highly experienced individuals from different industries. They explored how value is created in business ecosystems, focusing on both individual and organizational perspectives.
It was inspiring to have diverse industry representatives in the same room and to create a common understanding. Departing from economic value creation, the group extended its scope to ecological factors. An intense discussion arose about how environmental value can be created without negatively impacting economic value. Statements that economic value creation is still the predominant requirement were made, meaning that in many cases, even a slight reduction of economic value for the sake of ecological value would be treated with suspicion. As sustainability becomes increasingly relevant and regulations loom, the balance between economic and ecological value may shift in the near future.
Overall, the Smart Services for Sustainability – Circular Servitization discussion was thought-provoking and left participants eager to continue exploring the intersection of business and sustainability.
Spatial Data by Reik Leiterer
In a room buzzing with ideas, each data expert chimed into the discussion about the creation of a platform that would benefit cantons, individuals, and service providers. There was a shared understanding that it might not be possible to cater to everyone’s needs and that a simpler visualization and analytics approach may be the way forward. However, some uncertainties still remained, such as identifying where the necessary data is available and how it can be integrated, setting limits, and ensuring that data is not misinterpreted. Despite these challenges, the group remained enthusiastic about the potential benefits of the platform and is looking forward to overcoming these obstacles.
NLP & AI Technology by Lina Scarborough
The group opened the floor with how chatbots are great to answer questions, but what happens when users don’t know where to begin asking questions? This is a common issue in legal situations where the average client may not have the necessary background to understand what information is needed. Retrieval augmented language models like KATIE have emerged as a solution to this problem. These models use grounded reasoning and promote a chain of thought to handle complex queries and create a context for users who may not know what subset of questions to ask.
With the rise of machine-generated text, it’s becoming more difficult to distinguish between human and machine-generated content. While probabilistic token selection and frameworks like SCARECROW can help scrutinize machine-generated text, it can still be difficult, to nigh impossible, to identify. However, ChatGPTZero, an app that uses watermarking to create a statistical fingerprint in the sampling method, claims to be able to detect whether an essay is written by ChatGPT or a human – for instance, ChatGPT generally makes redundancy errors whereas humans make grammatical mistakes. This approach hopes to maintain the integrity of human-generated content in the face of increased machine-generated text.
The discussion then flowed into a lively and engaging presentation on how AI technology can make the tricky SQL “minefield” as easy to navigate as a soccer player scoring a goal – literally, by demonstrating SQL prompts on the soccer World Cup!
Smart Maintenance by Melanie Geiger
The five use case presentations highlighted the versatility of data technology in different applications, showcasing how it can be adapted to meet various needs. With input data ranging from domain knowledge to error log data, these use cases demonstrated how AI models can process and analyze complex data sets to provide valuable insights and decision support.
One of the key themes that emerged was the use of AI for diverse condition-based maintenance, specifically anomaly detection and fault diagnosis. By leveraging ML algorithms, these use cases were able to detect potential issues and predict equipment failures for timely maintenance and preventing downtime.
The highlight of the event was not only the apèro treats, but the opportunity to engage with the 60 participants and learn about their projects, challenges, solutions, and ideas for collaboration. Many attendees seemed to share this sentiment, as numerous participants were still engrossed in conversation at the end of the event, and some discussions had to be continued elsewhere. Those who wish to follow up on these conversations have the option to do so at SDS2023. On a more lowkey note, maybe you wanted to add someone on LinkedIn and send them a message. Here you go, this is your reminder!
Our conclusion of the event: the Alliance has many experts in various subtopics of data-driven value creation, but only together we can move faster.
We invite you to the second iteration of the Expert Day. Join us in an exchange of expertise and find inspiration. These following groups will participate:
Natural Language Processing & Big Data Technologies
Smart Maintenance
Smart Services
Spatial Data Analytics
Detailed Program:
15:00 – Welcome 15:30 – Keynote by Prof. Pierre Dersin 16:10 – Expert Group Meetings in breakout rooms (see below) 17:40 – Apéro
Data-driven Value Added for Words, Images and Things
Digital transformation is a defining feature of our epoch.
Abundance of data, immense increase in hardware processing capabilities and breakthroughs in analytics algorithms have made practical some of the visions put forward about three quarters of a century ago. The branch of Artificial Intelligence called Machine Learning, and in particular Deep Learning, permeates image processing, natural language processing ( “ words”) and smart maintenance (‘things’), and furthermore enables rich synergies between those three fields, which span a great deal of human activity, with profound potential impacts—some already visible, on industry, science, the arts and social life.
Natural Language Processing & Big Data Technologies
Everyone is talking about ChatGPT these days and some of its output is truly impressive! We will discuss how the most recent wave of text generation algorithms can transform business, science and teaching. The meeting will feature the following expert talks:
Grounded Copywriting with ChatGPT & Co by Michael Wechner (Wyona AG) + Colin Carter (Coop Rechtsschutz) Everyone talks about the pros and cons of ChatGPT, its competitors and how to combine the generated text with grounded knowledge. We will demonstrate how ChatGPT & Co can be applied in insurance and discuss the future of retrieval augmented language models.
Can we Identify Machine-Generated Text? An Overview of Current Approaches by Anastassia Shaitarova (UZH Institute for Computational Linguistics)
The detection of machine-generated text has become increasingly important due to the prevalence of automated content generation and its potential for misuse. In this talk, we will discuss the motivation for automatic detection of generated text. We will present the currently available methods, including feature-based classification as a “first line-of-defense.” We will provide an overview of the detection tools that have been made available so far and discuss their limitations. Finally, we will reflect on some open problems associated with the automatic discrimination of generated texts.
Using AI to Query the Football World Cup Database in Natural Language by Kurt Stockinger (ZHAW Institute for Applied Information Technology)
Football is one of the most popular sports on earth with millions of people watching the FIFA world cup. In this talk, we describe how we built a system to query the world cup database in natural language. We explain how we translate natural language into the database query language SQL using modern transformer architecture. We also demonstrate how we have used large language models such as Open AI’s GPT-3 and Google’s T5 to explain how the system interprets users’ questions.
We are looking forward to exchanging opinions, experiences and questions, and to exploring this exciting field together!
Smart Maintenance
The value of condition monitoring data: 5 use cases.
In this meeting of the Smart Maintenance Expert Group we will hear about successful student projects conducted together with industry partners from various fields. The focus points of the projects are very diverse, ranging from prediction of energy losses, through anomaly detection, fault diagnostics, prediction of the remaining useful life and optimal maintenance scheduling. We will have 5 short pitch presentations, followed by an interactive discussion of future interest topics of our expert group, including active feedback of all participants.
Anomaly Detection in Marine Engines with Convolutional Neural Networks (Company: WinGD)
Aircraft Scheduling Optimization based on Prognostics Degradation Models (Company: Swiss International Airlines)
Modeling Wake Energy Losses in Wind Farms using Graph Neural Networks (Company: Fluence Energy)
Using Error Code Patterns to Predict Service Requests on Production Machines with Machine Learning (Company: Zünd Systemtechnik).
Fault Detection in Solar Power Plants using Physics Informed Deep Learning (Company: Fluence Energy)
Smart services for sustainability – circular servitization
With data-driven services, industrial companies can create quantifiable value for their customers, partners and themselves. At the same time, these services also have the potential for ecological benefits, e.g., through optimized processes in operations or logistics. To make this possible, economic and ecological goals must be captured in a targeted and combined manner when designing the services.
The 1.5-hour workshop will discuss how specific problems from everyday business can be systematically addressed to create relevant added value for business and ecology. Participants will bring their own business issue and leave the workshop with a first approach on how to create economic and environmental value through smart services. The workshop will run through typical phases of a project in a compressed time format to give an impression of what such a project might look like on a larger scale.
Spatial Data Analytics
High-quality spatial data is increasingly available for free use. However, with the large amount of data and the sometimes very specific data types and formats, it is challenging to find the appropriate data sources. In addition, some of the data access platforms are only partially intuitive and can be used without expert knowledge. Accordingly, the question arises whether the full potential of the available data base could not be better exploited if data access and data sharing were simplified. In this co-creation workshop, concepts and approaches will be reflected and discussed with representatives from research and industry as well as from cantonal and federal agencies, with the aim of developing possible approaches for joint implementation.
Natural Language Processing (NLP) offers fundamental solutions for tasks such as text classification, automated chatbots, text summarization or speech recognition. How can the Insurance industry benefit from these AI technologies? This is the core question for the upcoming Expert Group Meeting “NLP in Insurance”, which will take place on Montag 31.10.2022 at 16:30 at ZHAW Lagerstrasse, Zurich.
Felix Müller, Senior Data Scientist at Mobiliar, will share insights into their usage of NLP for different applications (e.g. using transformers for claim handling). This will be followed be an open discussion among all participants (experts and users from academia and industry) and a nice apero.
The meeting is jointly organized by the NLP Expert Group, the Expert Group AI in Finance and Insurance, and the Swiss Association for Natural Language Processing (SwissNLP).
It is our pleasure to invite you to an upcoming event organised by the data innovation alliance and SwissNLP, the Swiss Association for Natural Language Processing (https://swissnlp.org/) on Tuesday, May 3rd at Nüü (Lagerstrasse 45, 8004 Zurich, ground floor).
We will combine an NLP Expert Group Meeting on Multimodal AI with the SwissNLP General Assembly 2022, followed by an apéro and socialising/networking.
16:00 – 17:00: SwissNLP General Assembly 2022
Annual and Financial Report 2021
Budget & activities for 2022
Board elections for 2022
Discussions & Varia
17:15 – 18:30: Expert Group Meeting on Multimodal AI An exciting new research direction, Multimodal Artificial Intelligence concerns the creation of AI models which can jointly process different types of inputs such as images, text, audio, video, or structured (tabular or graph) data. This opens up possibilities for applications which can interact with the world in new ways and address much more complex use cases.
A recent example is MUM [1], Google’s “Multitask Unified Model”, which can answer multimodal search queries such as “Are these hiking boots [see picture] suitable to hike Mount Fuji?”
In the meeting, we plan for 1-2 short presentations (10min each), followed by an open discussion of the topic. One potential (and intended) outcome is ideas for joint projects!
Looking for Speakers: If you would like to share your experience in Multimodal AI with the group, please let us know. The presentation may include a success story, a collection of best practices, an open problem, or even an outline of a project that would benefit from Multimodal AI. It would be great to discuss positive experiences, as well as blockers and limitations. If you would like to present, please send us your presentation titles (email to ciel@zhaw.ch, natalia.korchagina@pwc.ch). Deadline is Monday, April 25.
From 18:30: Apéro & socialising/networking Join us to exchange & mingle over drinks & snacks!
It would be great to see many of you in person, but online participation is also possible for both events.
Please register using the link from the invitation mail. Deadline for registration: Wednesday, April 27.
SwissText is an annual conference that brings together text analytics experts from industry and academia. It is organised by the Swiss Association for Natural Language Processing (SwissNLP) in collaboration with the University of Applied Sciences and Arts of Southern Switzerland (SUPSI), the Data Innovation Alliance as well as the Zurich University of Applied Sciences (ZHAW).
For registration please use this link: REGISTRATION
Topic: Swiss German Speech Processing Date: Tuesday, November 9, 2021 Time: 17:00 Location: online or on premise in Zurich (will be decided based on preference of participants) Registration: until November 1st
It has become rather difficult over the last two years for professionals to meet and informally discuss topics in their fields. Databooster is about bringing research and industry together so that new innovative projects in different data-intensive fields can be born. But how can we stimulate this exchange? The NLP Focus Group of the Databooster initiative (with Focus Group Leaders Mark Cieliebak, Philipp Thomann, Natalia Korchagina) addressed this challenge by organizing the first “Battle of NLP Ideas” workshop at the SwissText conference. The event was held on June 15, 2021 online.
About 50 NLP enthusiasts joined our virtual battle. The goal of this interactive session was to give stage to as many ideas as possible, multiplying the chances for the people with similar ideas to meet and possibly form an alliance. The final prize of the battle was a chance to turn your idea into an Innosuisse-funded project!
Our battle had several rounds. In the first round, the participants were split in small groups of 3 to 5 where everyone could pitch their idea to other member of the group in 10-20 seconds, and the most promising idea of each small group was selected by voting. In the next round, a small presentation of an idea was prepared by each group and then pitched in front of the members of other groups. The most promising ideas were discussed in more details with all participants. After three rounds and the final vote, we had our winning ideas:
• Fake-News detection • Topic Segmentation and Classification in Multi-topic Reports • Automatic Contract Generation
It has been an inspiring couple of hours. There is so much research potential and industry need in working NLP products in different domains, from journalism to ecology. To assist our winners in bringing their ideas to life, the focus group leaders organized follow-up meetings where the ideas were further developed, and their market potential was discussed in more details.
Fake News Detection API, capable to recognise fake news, has proven to be in huge demand. Three large Swiss media companies and two Swiss universities (FHNW and ZHAW) are discussing to collaborate. The plan is to start with an Innocheck which will be most likely followed by an Innosuisse application.
A first follow-up workshop on the Topic Segmentation idea showed that there is large interest from academia, but only limited significance to industry. Automatic contract generation did not receive enough support from the industrial partners after the first follow-up.
We wish all our participants to turn their ideas into real projects and keep our fingers crossed for the winners!
The next battle of NLP ideas will be fought at the upcoming SwissText conference in June2022. We are looking forward to seeing many fellow NLP enthusiasts joining the contest.