The NeSy workshop series celebrates the integration of neural and symbolic thinking, technologies, theories and techniques of Artificial Intelligence and Machine Learning.
NeSy is the annual meeting of the Neural-Symbolic Learning and Reasoning Association (see also www.neural-symbolic.org).
Neural networks and statistical Machine Learning have obtained industrial relevance in a number of areas from retail to healthcare, achieving state-of-the-art performance at language modelling, speech recognition, graph analytics, image, video and sensor data analysis. Symbolic AI, on the other hand, is challenged by such unstructured data, but is recognised as being in principle transparent, in that reasoned facts from knowledge-bases can be inspected to interpret how decisions follow from input. Neural and symbolic methods also contrast in the problems that they excel at: scene recognition from images appears to be a problem still outside the capabilities of symbolic systems, for example, while neural networks are not yet sufficient for industrial-strength complex planning scenarios and deductive reasoning tasks.
Neurosymbolic AI aims to build rich computational models and systems by combining neural and symbolic learning and reasoning paradigms. This combination hopes to form synergies among their strengths while overcoming their complementary weaknesses.
The NeSy workshop series is the premier venue for the presentation and discussion of the theory and practice of neural-symbolic computing systems. Since 2005, NeSy has provided an atmosphere for the free exchange of ideas bringing together the community of scientists and practitioners that straddle the line between deep learning and symbolic AI.
IJCLR Keynotes, NeSy Tutorial and Invited Speakers:
Zhi-Hua Zhou, Nanjing University
Josh Tenenbaum, MIT
Francesca Rossi, IBM Research
Gary Marcus, RobustAI
Hector Geffner, Universitat Pompeu Fabra
Francesca Toni, Imperial College London
Guy Van den Broeck, UCLA
Zachary Lipton, CMU
Ian Horrocks, University of Oxford
Hava Seigelmann, University of Massachusetts, Amherst
Craig Atkinson, University of Otago
Razvan Pascanu, Google DeepMind
Rahaf Aljundi, Toyota Motor Europe
Tom Mitchell, CMU
Bing Liu, University of Illinois, Chicago
Adam White, City, University of London
Artur d'Avila Garcez, City, University of London
Pascal Hitzler, Kansas State University
Aaron Eberhart, Kansas State University
Monireh Ebrahimi, lBM Watson San Francisco
Federico Bianchi, Bocconi University, Milan
Michael Spranger, Sony AI, Tokyo
Dan Philps, Rothko investments, London
Joe Townsend, Fujitsu, London
Alex Jaimes, Dataminr, New York
Dragos Margineantu, Boeing R&D, Seattle
WORKSHOP SCHEDULE - times shown are BST (British Summer Time)
Mon, 25th Oct 2021
9:40 Welcome - Luc De Raedt, Nikos Katzouris, Ute Schmid, Artur Garcez, Sebastijan Dumancic.
10:00 IJCLR Keynote - Zhi-Hua Zhou, Nanjing University, China.
11:00 NeSy and Continual Learning (chair: Natalia Diaz Rodriguez, ENSTA, Paris, France)
11:00 Hava Seigelmann, University of Massachusetts, Amherst, USA. Lifelong Learning: The Cutting Edge of Artificial Intelligence.
11:20 Craig Atkinson, University of Otago, New Zealand. Achieving Continual Learning with Pseudo-Rehearsal.
11:40 Razvan Pascanu, Google DeepMind, London, UK. What Should We Expect from a Lifelong Learning System?
12:30 IJCLR Journal paper presentations:
12:30 Gustav Sourek, Filip Zelezny, Ondrej Kuzelka. Beyond Graph Neural Networks with Lifted Relational Neural Networks.
12:50 Varun Embar, Sriram Srinivasan, Lise Getoor. A Comparison of Statistical Relational Learning and Graph Neural Networks for Aggregate Graph Queries.
13:10 Tirtharaj Dash, Ashwin Srinivasan, A Baskar. Inclusion of Domain-Knowledge into GNNs using Mode-Directed Inverse Entailment.
14:00 Joint poster session 1 (list of NeSy papers below)
15:30 NeSy tutorials (parallel sessions)
15:30 - 16:30 Tutorial 1: Deep Deductive Reasoning over Semantic Web Logics, Pascal Hitzler, Aaron Eberhart, Kansas State University, USA, Monireh Ebrahimi, lBM Watson San Francisco, USA, Federico Bianchi, Bocconi University, Milan, Italy.
15:30 - 16:30 Tutorial 2: NeSy and Continual Learning (Chair: Danny Silver, Acadia University, Canada)
15:30 The Challenges and Settings for Continual Learning, Rahaf Aljundi, Toyota Motor Europe, Belgium.
15:50 Lessons from the Never Ending Language Learner, Tom Mitchell, CMU, USA.
16:10 Continual Learning for Knowledge Accumulation, Bing Liu, University of Illinois, Chicago, USA.
15:30 - 16:30 Tutorial 3: NeSy and Explainable AI, Adam White, Artur Garcez, City, University of London, UK.
16:45 IJCLR Keynote - Josh Tenenbaum, MIT, USA.
17:30 End of Day 1
Tue, 26th Oct 2021
10:00 IJCLR Keynote - Hector Geffner, Universitat Pompeu Fabra, Barcelona, Spain.
11:00 NeSy and the Semantic Web (chairs: Dagmar Groman, University of Vienna, Luis Espinosa Anke, Cardiff University, Thierry Declerck, DFKI, Ernesto Jimenez-Ruiz, City, University of London)
11:00 Invited talk - Ian Horrocks, University of Oxford, UK.
11:30 Mirantha Jayathilaka, Tingting Mu and Uli Sattler Ontology-based n-ball Concept Embeddings Informing Few-shot Image Classification
11:45 Discussion (Moderator: Dagmar Groman, University of Vienna, Austria)
12:30 IJCLR Journal paper presentations:
14:00 Joint poster session 2 (list of NeSy papers below)
15:30 NeSy session (chair: Luis Lamb, Federal University of Rio Grande do Sul, Brazil)
15:30 Paolo Dragone, Stefano Teso and Andrea Passerini. Neuro-Symbolic Constraint Programming for Structured Prediction.
15:45 Harald Strömfelt, Luke Dickens, Artur d'Avila Garcez and Alessandra Russo. Coherent and Consistent Relational Transfer Learning with Autoencoders.
16:00 Invited Talk - Guy Van den Broeck, UCLA, USA. Reasoning about Learned Models' Behaviour
16:45 IJCLR Keynote - Gary Marcus, RobustAI, CA, USA.
17:30 End of Day 2
Wed, 27th Oct 2021
10:00 IJCLR Keynote - Francesca Toni, Imperial College London, UK.
11:00 NeSy industry session (chair: Artur Garcez, City, University of London, UK)
11:00 Michael Spranger, Sony AI, Tokyo, Japan
11:10 Dan Philps, Rothko investments, London, UK. Death by XAI in Finance.
11:20 Joe Townsend, Fujitsu, London, UK. Towards Trusted AI: Rule-based explanations of CNN behaviour.
11:30 Alex Jaimes, Dataminr, New York, USA
11:40 Dragos Margineantu, Boeing R&D, Seattle, USA
12:30 NeSy and XAI (chairs: Freddy Lecue, Thales Montreal, Canada, Veronika Thost, IBM Research Zurich, Switzerland)
12:30 Theodoros Kasioumis, Joe Townsend and Hiroya Inakoshi Elite BackProp: Training Sparse Interpretable Neurons.
12:45 Anna Himmelhuber, Sonja Zillner, Stephan Grimm, Martin Ringsquandl, Mitchell Joblin and Thomas Runkler. A New Concept for Explaining Graph Neural Networks.
13:00 Invited Talk - Zachary Lipton, CMU, USA. Integrating Symbolic Structure & Statistical Learning to Adapt to Distribution Shift.
14:00 IJCLR Keynote - Francesca Rossi, IBM Watson Research Centre, New York, USA.
15:00 IJCLR Panel Discussion - Future Challenges in Learning and Reasoning (Moderator: Stephen Muggleton, Imperial College London, UK)
16:30 IJCLR Community Meeting (chairs: Luc De Raedt, KU Leuven, Belgium, Nikos Katzouris, NCSR Demokritos, Greece)
Poster session 1: Mon 25th Oct 2021, 14:00 - 15:15h BST
Paolo Dragone, Stefano Teso and Andrea Passerini. Neuro-Symbolic Constraint Programming for Structured Prediction.
Patrick Betz, Mathias Niepert, Pasquale Minervini and Heiner Stuckenschmidt. Backpropagating through Markov Logic Networks.
Harald Strömfelt, Luke Dickens, Artur d'Avila Garcez and Alessandra Russo. Coherent and Consistent Relational Transfer Learning with Auto-encoders.
Thomas Winters, Giuseppe Marra, Robin Manhaeve and Luc De Raedt. DeepStochLog: Neural Stochastic Logic Programming.
Jinyung Hong and Theodore Pavlic. An Insect-Inspired Randomly Weighted Neural Network with Random Fourier Features For Neuro-Symbolic Relational Learning.
Pasquale Minervini, Daniel Daza, Erik Arakelyan and Michael Cochez. Complex Query Answering with Neural Link Predictors.
Justin Brody, Bobby Austin, Omar Khater, Christopher J. Maxey, Matthew D Goldberg, Timothy Clausner, Darsana Josyula and Don Perlis. Using Neural Networks to Control Glut in the Active Logic Machine.
Daniel Silver and Ahmed Galila. Learning Arithmetic from Handwritten Images with the Aid of Symbols.
Poster session 2: Tue 26th Oct 2021, 14:00 - 15:15h BST
Mirantha Jayathilaka, Tingting Mu and Uli Sattler. Ontology-based n-ball Concept Embeddings Informing Few-shot Image Classification.
Theodoros Kasioumis, Joe Townsend and Hiroya Inakoshi. Elite BackProp: Training Sparse Interpretable Neurons.
Pedro Zuidberg Dos Martires. Neural Semirings.
Anna Himmelhuber, Sonja Zillner, Stephan Grimm, Martin Ringsquandl, Mitchell Joblin and Thomas Runkler. A New Concept for Explaining Graph Neural Networks.
Rohan Deshpande, Jerry Chen and Isabelle Lee. RecT: A Recursive Transformer Architecture for Generalizable Mathematical Reasoning.
Nuri Cingillioglu and Alessandra Russo. pix2rule: End-to-end Neuro-symbolic Rule Learning.
Edgar Altszyler, Pablo Brusco, Nikoletta Basiou, John Byrnes and Dimitra Vergyri. Zero-shot Multi-Domain Dialog State Tracking Using Prescriptive Rules.
Samy Badreddine and Michael Spranger. Real Logic with Aggregate Functions.
Se-In Jang and Alexandre Thiery. Explainable Diabetic Retinopathy Classification Based on Neural-Symbolic Learning.
Call For Journal Track Submissions - ML journal
Deadlines: May 2020, Sep 2020, Dec 2020, Mar 2021, Jun 2021, Dec 2021, etc.
NeSy invites theoretical and applied paper submissions to the IJCLR journal track combining deep learning and symbolic AI. We further invite paper submissions detailing experimental and in-the-wild systems, and papers on topics where neural-symbolic computing has a strong use case. Topics of interest include, but are not limited to:
Knowledge representation and reasoning using deep neural networks
Symbolic knowledge extraction from neural and statistical learning systems
Explainable AI models, systems and techniques that integrate connectionist and symbolic paradigms
Neural-symbolic cognitive models
Biologically-inspired neural-symbolic integration
Continual learning, integration of logic and probabilities with neural networks
Neural-symbolic methods for structured learning, including transfer, meta, multi-task and relational learning
Novel connectionist systems able to perform traditionally symbolic AI tasks (e.g. logical deduction)
Novel symbolic systems able to perform traditionally connectionist tasks (e.g. unstructured data analysis)
Applications of neural-symbolic and hybrid systems, including simulation, finance, robotics, semantic web, software engineering, systems engineering, bioinformatics and visual intelligence.
Authors of accepted papers will be invited to present their work at the NeSy'20/21 Workshop.
Call for Workshop Paper Submissions
In addition to the journal track, NeSy invites submissions of the latest and ongoing research work for presentation at the workshop. Research papers in any of the above areas of neural-symbolic computing are welcome as well as submissions targeting one of the workshop special sessions:
Semantic Web and Deep Learning
NeSy and Continual Learning
NeSy for Explainable AI
Industry experience and application track
All accepted papers will be published by CEUR and are expected to be presented at the workshop. Revised and extended versions of the best papers will be invited for submission to the NeSy journal track. A selection of the accepted papers will be chosen for poster presentation.
Researchers and practitioners are invited to submit original papers that have not been submitted for review or published elsewhere. Submitted papers need not be anonymous, must be written in English, should be formatted using single column and 11pt font, and should not exceed 8 pages in the case of research and experience papers, or 4 pages in the case of position papers (including all figures, but excluding references and appendices). You are encouraged to use the CEUR Latex template or the CEUR Word template.
All submitted papers will be judged based on their relevance, originality, significance, technical quality and organisation.
Deadline for paper submission: 7 July 2021
30 June 2021(23:59 Anywhere on Earth)
Notification of paper acceptance: 30 August 2021 (23:59 Anywhere on Earth)
Camera-ready paper due: 13 September 2021 (23:59 Anywhere on Earth)
NeSy Workshop dates: 25-27 October 2021
2020/21 Workshop Organizers
Artur d'Avila Garcez, City, University of London, UK
Ernesto Jimenez-Ruiz, City, University of London, UK
Natalia Diaz Rodriguez, ENSTA ParisTech, FR
Dagmar Gromann, University of Vienna, AT
Freddy Lecue, INRIA and Thales, Montreal, CA
Danny Silver, Acadia University, Canada
Luis Lamb, University of Rio Grande do Sul, Brazil
Veronika Thost, IBM Research Zurich, Switzerland
NeSy History and Past Proceedings
NeSy'17 @ City, University of London
Asan Agibetov, Medical University of Vienna, AT
Elvira Amador-Domínguez, Univ. Politécnica de Madrid, Spain
Vito Walter Anelli, Polytechnic University of Bari, Italy
Jiaoyan Chen, University of Oxford, UK
Bernardo Cuenca Grau, University of Oxford, UK
Vincenzo Cutrona, University of Milano - Bicocca, IT
Ivan Donadello, University of Bolzano, IT
Derek Doran, Wright State University, OH, USA
Vasilis Efthymiou, ICS-FORTH, Greece
Eleonora Giunchiglia, University of Oxford, UK
Frank Van Harmelen, Vrije Universiteit Amsterdam, NL
Pascal Hitzler, Kansas State University, USA
Steffen Hölldobler, TU Dresden, DE
Andreas Holzinger, Medical University Graz, AT
Kristian Kersting, TU Darmstadt, DE
Luis Lamb, University of Rio Grande do Sul, Brazil
Freddy Lecue, INRIA and Thales, Montreal, CA
Thomas Lukasiewicz, University of Oxford, UK
Monireh Ebrahimi, Wright State University, USA
Carlos Maestre, Valeo, FR
Pasquale Minervini, University College London, UK
Summaya Mumtaz, University of Oslo, NO
Erik Bryhn Myklebust, NIVA, NO
Heiko Paulheim, University of Mannheim, Germany
Catia Pesquita, Universidade de Lisboa, PT
Alina Petrova, University of Oxford, UK
Kamruzzaman Sarker, Kansas State University, USA
Luciano Serafini, Fondazione Bruno Kessler, Trento, IT
Michael Spranger, Sony CSL and Sony AI, Tokyo, JP
Kavitha Srinivas, IBM Research, USA
Andreas Theodorou, Umeå University, SE
The workshop will include specialised tutorials and extra time for audience discussion, allowing the group to obtain a better understanding of the issues, challenges and ideas being presented. Authors of accepted papers may be assigned either an oral or poster presentation slot in the final workshop programme. Accepted workshop track papers will be published by CEUR. Accepted journal track papers will be published by Springer. Registration is open to anybody willing to participate.