Aarne Talman


I'm a language technology researcher working on language understanding and reasoning

     

Learn more

About me


I work as a Lead AI Scientist at Silo AI. I'm also a Visiting Scholar in Language Technology at University of Helsinki. I have almost 20 years of experience in research, software engineering, consulting and leadership.

Research


My research focuses on natural language understanding, reasoning and natural language inference using machine learning. I’m interested in what understanding language consists of, how it can be modelled in AI and how these models can be evaluated.

Natural Language Understanding

My research focuses on reasoning and natural language inference. I have developed new methods and models as well as studied the limitations of various benchmarks and datasets in NLU.

Speech Recognition and Prosody

I have done research in speech processing, including automatic speech recognition and prosody. I've developed speech recognition models used in production by thousands of end users.

Neural Machine Translation

I have developed machine translation models and state-of-the-art neural machine translation systems used in production by thousands of end users.

Papers and Talks


Peer-Reviewed Papers

  1. Aarne Talman, Hande Celikkanat, Sami Virpioja, Markus Heinonen, Jörg Tiedemann. 2023. Uncertainty-Aware Natural Language Inference with Stochastic Weight Averaging. Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa). [bibtex] [pdf] [code]
  2. Aarne Talman, Marianna Apidianaki, Stergios Chatzikyriakidis, Jörg Tiedemann. 2022. How Does Data Corruption Affect Natural Language Understanding Models? A Study on GLUE datasets. Proceedings of the 11th Joint Conference on Lexical and Computational Semantics. [bibtex] [pdf] [data and code]
  3. Aarne Talman, Marianna Apidianaki, Stergios Chatzikyriakidis, Jörg Tiedemann. 2021. NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance. Proceedings of the 23rd Nordic Conference on Computational Linguistics. [bibtex] [pdf] [data and code]
  4. Aarne Talman, Antti Suni, Hande Celikkanat, Sofoklis Kakouros, Jörg Tiedemann and Martti Vainio. 2019. Predicting Prosodic Prominence from Text with Pre-trained Contextualized Word Representations. Proceedings of the 22nd Nordic Conference on Computational Linguistics. [bibtex] [pdf] [corpus and code]
  5. Aarne Talman, Umut Sulubacak, Raúl Vázquez, Yves Scherrer, Sami Virpioja, Alessandro Raganato, Arvi Hurskainen, and Jörg Tiedemann. 2019. The University of Helsinki submissions to the WMT19 news translation task. Proceedings of the Fourth Conference on Machine Translation: Shared Task Papers. [bibtex] [pdf]
  6. Aarne Talman and Stergios Chatzikyriakidis. 2019. Testing the Generalization Power of Neural Network Models Across NLI Benchmarks. Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP. [bibtex] [pdf]
  7. Aarne Talman, Anssi Yli-Jyrä and Jörg Tiedemann. 2019. Sentence Embeddings in NLI with Iterative Refinement Encoders. Natural Language Engineering 25(4). [bibtex] [pdf] [code]

Theses

  1. Aarne Talman. 2006. Path Grammars and the Generative Capacity of Dynamic Syntax. Master's Dissertation. King's College London.
  2. Aarne Talman. 2005. A Limit on Artificial Intelligence? – The Gödelian Case. Bachelor's Thesis. The London School fo Economics and Political Science.

Talks

  1. How Does Data Corruption Affect Natural Language Understanding Models?. 29th September 2022, Research Seminar in Language Technology, University of Helsinki. [pdf]
  2. How Does Data Corruption Affect Natural Language Understanding Models? A Study on GLUE datasets. 15th July 2022, The 11th Joint Conference on Lexical and Computational Semantics (*SEM) 2022, Seattle, Washington, USA.
  3. NLI Data Sanity Check: Assessing the Effect of Data Corruption on Model Performance. 2 June 2021, The 23rd Nordic Conference on Computational Linguistics, Reykjavik. [pdf]
  4. Predicting Prosodic Prominence from Text with Pre-trained Contextualized Word Representations. 14 November 2019, Research Seminar in Language Technology, University of Helsinki. [pdf]
  5. Predicting Prosodic Prominence from Text with Pre-trained Contextualized Word Representations. 2 October 2019, The 22nd Nordic Conference on Computational Linguistics, Turku. [pdf]
  6. Neural Network models of NLI fail to capture the general notion of inference, 8 March 2019, CLASP Seminar, University of Gothenburg. [pdf]
  7. State-of-the-Art Natural Language Inference Systems Fail to Capture the Semantics of Inference, 25 October 2018, Research Seminar in Language Technology, University of Helsinki. [pdf]
  8. Natural Language Inference with Hierarchical BiLSTM’s, 28 September 2018, FoTran 2018. [pdf]
  9. Natural Language Inference - Another Triumph for Deep Learning?, 23 November 2017, Research Seminar in Language Technology, University of Helsinki. [pdf]

Teaching


University of Helsinki

Other

Software and Data


Visit my GitHub profile for a more complete collection of software and data.

Software

  1. Uncertainty-Aware NLI with SWAG: Code for our 2023 NoDaLiDa paper.
  2. NLU Dataset Diagnostics: Scripts for our 2022 *SEM paper.
  3. NLI Data Sanity Check: Data and scripts for our 2021 NoDaLiDa paper.
  4. Prosody: A system for predicting prosodic prominence from written text.
  5. Natural Language Inference: Natural language inference system written in Python and PyTorch implementing the HBMP sentence encoder.

Data

  1. Helsinki Prosody Corpus: The prosody corpus contains automatically generated, high quality prosodic annotations for the LibriTTS corpus (Zen et al. 2019) using the Continuous Wavelet Transform Annotation method (Suni et al. 2017).
    • Language: English
    • License: CC BY 4.0
    • Paper

CV


Download my full CV here.

Education

Employment

  • 2023 - present, Lead AI Scientist, Silo AI
  • 2022 - present, Visiting Scholar, University of Helsinki
    Working on natural language understanding.
  • 2023 - 2023, Senior Manager, Accenture
  • 2022 - 2022, Lead AI Engineer, Silo AI
    Working on natural language processing and search.
  • 2021 - 2022, Senior AI Engineer, Silo AI
    Working on natural language processing and search.
  • 2018 - present, Doctoral Candidate, Language Technology, University of Helsinki
    Working on computational semantics and natural language processing.
  • 2019 - present, Founder & CEO, Basement AI
    Basement AI is a Nordic artificial intelligence research lab and consulting company specializing in natural language processing and machine learning.
  • 2020 - 2021, UK CTO, Nordcloud
    Nordcloud is a leading public cloud professional and managed services company. Lead- ing a team of architects and engineers.
  • 2015 - 2018, Associate Director, Consulting, Gartner.
  • 2012 - 2015, Consultant, Accenture.
  • 2011 - 2012, Research Student, London School of Economics.
  • 2009 - 2011, Product Manager, Nokia.
  • 2008 - 2009, Manager, Nokia.
  • 2006 - 2008, Systems Analyst, Tieto.
  • 2006 - 2006 (2 months), Software Developer, Valuatum.

Get In Touch!


Contact me by email or on social media!