-
ICML 2023 Topological Deep Learning Challenge : Design and Results
Authors:
Mathilde Papillon,
Mustafa Hajij,
Helen Jenne,
Johan Mathe,
Audun Myers,
Theodore Papamarkou,
Tolga Birdal,
Tamal Dey,
Tim Doster,
Tegan Emerson,
Gurusankar Gopalakrishnan,
Devendra Govil,
Aldo Guzmán-Sáenz,
Henry Kvinge,
Neal Livesay,
Soham Mukherjee,
Shreyas N. Samaga,
Karthikeyan Natesan Ramamurthy,
Maneel Reddy Karri,
Paul Rosen,
Sophia Sanborn,
Robin Walters,
Jens Agerberg,
Sadrodin Barikbin,
Claudio Battiloro
, et al. (31 additional authors not shown)
Abstract:
This paper presents the computational challenge on topological deep learning that was hosted within the ICML 2023 Workshop on Topology and Geometry in Machine Learning. The competition asked participants to provide open-source implementations of topological neural networks from the literature by contributing to the python packages TopoNetX (data processing) and TopoModelX (deep learning). The chal…
▽ More
This paper presents the computational challenge on topological deep learning that was hosted within the ICML 2023 Workshop on Topology and Geometry in Machine Learning. The competition asked participants to provide open-source implementations of topological neural networks from the literature by contributing to the python packages TopoNetX (data processing) and TopoModelX (deep learning). The challenge attracted twenty-eight qualifying submissions in its two-month duration. This paper describes the design of the challenge and summarizes its main findings.
△ Less
Submitted 18 January, 2024; v1 submitted 26 September, 2023;
originally announced September 2023.
-
Neural incomplete factorization: learning preconditioners for the conjugate gradient method
Authors:
Paul Häusner,
Ozan Öktem,
Jens Sjölund
Abstract:
Finding suitable preconditioners to accelerate iterative solution methods, such as the conjugate gradient method, is an active area of research. In this paper, we develop a computationally efficient data-driven approach to replace the typically hand-engineered algorithms with neural networks. Optimizing the condition number of the linear system directly is computationally infeasible. Instead, our…
▽ More
Finding suitable preconditioners to accelerate iterative solution methods, such as the conjugate gradient method, is an active area of research. In this paper, we develop a computationally efficient data-driven approach to replace the typically hand-engineered algorithms with neural networks. Optimizing the condition number of the linear system directly is computationally infeasible. Instead, our method generates an incomplete factorization of the matrix and is, therefore, referred to as neural incomplete factorization (NeuralIF). For efficient training, we utilize a stochastic approximation of the Frobenius loss which only requires matrix-vector multiplications. At the core of our method is a novel messagepassing block, inspired by sparse matrix theory, that aligns with the objective of finding a sparse factorization of the matrix. By replacing conventional preconditioners used within the conjugate gradient method by data-driven models based on graph neural networks, we accelerate the iterative solving procedure. We evaluate our proposed method on both a synthetic and a real-world problem arising from scientific computing and show its ability to reduce the solving time while remaining computationally efficient.
△ Less
Submitted 5 February, 2024; v1 submitted 25 May, 2023;
originally announced May 2023.
-
Dark Patterns in the Interaction with Cookie Banners
Authors:
Philip Hausner,
Michael Gertz
Abstract:
Dark patterns are interface designs that nudge users towards behavior that is against their best interests. Since humans are often not even aware that they are influenced by these malicious patterns, research has to identify ways to protect web users against them. One approach to this is the automatic detection of dark patterns which enables the development of tools that are able to protect users…
▽ More
Dark patterns are interface designs that nudge users towards behavior that is against their best interests. Since humans are often not even aware that they are influenced by these malicious patterns, research has to identify ways to protect web users against them. One approach to this is the automatic detection of dark patterns which enables the development of tools that are able to protect users by proactively warning them in cases where they face a dark pattern. In this paper, we present ongoing work in the direction of automatic detection of dark patterns, and outline an example to detect malicious patterns within the domain of cookie banners.
△ Less
Submitted 27 March, 2021;
originally announced March 2021.
-
Very-high-energy gamma-rays from the Universe's middle age: detection of the z=0.940 blazar PKS 1441+25 with MAGIC
Authors:
MAGIC Collaboration,
M. L. Ahnen,
S. Ansoldi,
A. Antonelli,
P. Antoranz,
A. Babic,
B. Banerjee,
P. Bangale,
U. Barres de Almeida,
J. A. Barrio,
W. Bednarek,
E. Bernardini,
B. Biassuzzi,
A. Biland,
O. Blanch,
S. Bonnefoy,
G. Bonnoli,
F. Borracci,
T. Bretz,
E. Carmona,
A. Carosi,
A. Chatterjee,
R. Clavero,
P. Colin,
E. Colombo
, et al. (229 additional authors not shown)
Abstract:
The flat-spectrum radio quasar PKS 1441+25 at a redshift of z = 0.940 is detected between 40 and 250 GeV with a significance of 25.5 σ using the MAGIC telescopes. Together with the gravitationally lensed blazar QSO B0218+357 (z = 0.944), PKS 1441+25 is the most distant very high energy (VHE) blazar detected to date. The observations were triggered by an outburst in 2015 April seen at GeV energies…
▽ More
The flat-spectrum radio quasar PKS 1441+25 at a redshift of z = 0.940 is detected between 40 and 250 GeV with a significance of 25.5 σ using the MAGIC telescopes. Together with the gravitationally lensed blazar QSO B0218+357 (z = 0.944), PKS 1441+25 is the most distant very high energy (VHE) blazar detected to date. The observations were triggered by an outburst in 2015 April seen at GeV energies with the Large Area Telescope on board Fermi. Multi-wavelength observations suggest a subdivision of the high state into two distinct flux states. In the band covered by MAGIC, the variability time scale is estimated to be 6.4 +/- 1.9 days. Modeling the broadband spectral energy distribution with an external Compton model, the location of the emitting region is understood as originating in the jet outside the broad line region (BLR) during the period of high activity, while being partially within the BLR during the period of low (typical) activity. The observed VHE spectrum during the highest activity is used to probe the extragalactic background light at an unprecedented distance scale for ground-based gamma-ray astronomy.
△ Less
Submitted 12 January, 2018; v1 submitted 14 December, 2015;
originally announced December 2015.
-
The 2013 multiwavelength campaign on the Narrow-Line Seyfert 1 galaxy 1H 0323+342: a rosetta stone for the jet/disk paradigm
Authors:
Omar Tibolla,
Sarah Kaufmann,
Luigi Foschini,
Karl Mannheim,
Shu Zhang,
Jian Li,
Emmanouil Angelakis,
Lars Fuhrmann,
Paul Haeusner,
Jannik Kania,
Dominik Elsaesser,
Annika Kreikenbohm,
Robert Schulz,
Matthias Kadler
Abstract:
Narrow-Line Seyfert 1 galaxies have been established as a new class of gamma-ray emitting AGN with relatively low black hole masses, but near-Eddington accretion rates. Other extragalactic gamma-ray sources observed so far such as Flat Spectrum Radio Quasars, Radio Galaxies, and BL Lacertae Objects generally exhibit much higher black hole masses and, in the case of BL Lac objects and FRI Radio Gal…
▽ More
Narrow-Line Seyfert 1 galaxies have been established as a new class of gamma-ray emitting AGN with relatively low black hole masses, but near-Eddington accretion rates. Other extragalactic gamma-ray sources observed so far such as Flat Spectrum Radio Quasars, Radio Galaxies, and BL Lacertae Objects generally exhibit much higher black hole masses and, in the case of BL Lac objects and FRI Radio Galaxies, much lower accretions rates. The multifrequency campaign of 2013 centered on the bright source 1H 0323+342 will provide further insights into the nature of the jets and their gamma ray production mechanisms in a largely unexplored corner of AGN parameter space. Here, we show preliminary results of this campaign and discuss them.
△ Less
Submitted 17 June, 2013;
originally announced June 2013.
