TBTin Barisin

Tin Barisin

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Tin Barisin has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Toward Generalized Entropic Sparsification for Convolutional Neural Networks

    by Tin Barisin, Illia Horenko

    Convolutional neural networks (CNNs) are reported to be overparametrized. The search for optimal (minimal) and sufficient architecture is an NP-hard problem: if the network has N neurons, then there are 2N possibilities to connect them—and therefore 2N possible architectures and 2N Boolean hyperparameters to encode them. Selecting the best possible hyperparameter out of them becomes an Np-hard problem since 2N grows in N faster then any polynomial Np. Here, we introduce a layer-by-layer data-driven pruning method based on the mathematical idea aiming at a computationally scalable entropic relaxation of the pruning problem. The sparse subnetwork is found from the pretrained (full) CNN using the network entropy minimization as a sparsity constraint. This allows deploying a numerically scalable algorithm with a sublinear scaling cost. The method is validated on several benchmarks (architectures): on MNIST (LeNet), resulting in sparsity of 55% to 84% and loss in accuracy of just 0.1% to 0.5%, and on CIFAR-10 (VGG-16, ResNet18), resulting in sparsity of 73% to 89% and loss in accuracy of 0.1% to 0.5%.


  • Simulation of Microstructures and Machine Learning

    by Katja Schladitz, Claudia Redenbach, Tin Barisin, Christian Jung, Natascha Jeziorski, Lovro Bosnar, Juraj Fulir, Petra Gospodnetić


  • Open Access

    Riesz Networks: Scale-Invariant Neural Networks in a Single Forward Pass

    by Tin Barisin, Katja Schladitz, Claudia Redenbach

    Scale invariance of an algorithm refers to its ability to treat objects equally independently of their size. For neural networks, scale invariance is typically achieved by data augmentation. However, when presented with a scale far outside the range covered by the training set, neural networks may fail to generalize. Here, we introduce the Riesz network, a novel scale- invariant neural network. Instead of standard 2d or 3d convolutions for combining spatial information, the Riesz network is based on the Riesz transform which is a scale-equivariant operation. As a consequence, this network naturally generalizes to unseen or even arbitrary scales in a single forward pass. As an application example, we consider detecting and segmenting cracks in tomographic images of concrete. In this context, ‘scale’ refers to the crack thickness which may vary strongly even within the same sample. To prove its scale invariance, the Riesz network is trained on one fixed crack width. We then validate its performance in segmenting simulated and real tomographic images featuring a wide range of crack widths. An additional experiment is carried out on the MNIST Large Scale data set.


  • Open Access

    Analysis of the bond behavior of a GFRP rebar in concrete by in-situ 3D imaging test

    by Szymon Grzesiak, Tin Barisin, Katja Schladitz, Matthias Pahn

    Computed tomography combined with mechanical tests offers completely new insight into the behavior of concrete samples under stress. Particularly the development of new fiber reinforcement materials for concrete elements requires appropriate material models and thus for investigating the interior of the concrete structure. In 3D image data obtained by computed tomography, local structural changes within the sample due to mechanical loading can be observed without further altering the sample. We applied this state-of-the-art approach to a concrete core with an embedded glass fiber reinforced polymer rebar under increasing forces applied to pull out the rebar. In this paper, authors describe a novel in-situ setup for non-destructively 3D imaging during the pull-out test. Conducting the pull-out test leads to the formation of local pore volume changes along the rebar. These pore volume changes are not only visualized but quantified analytically based on the images. Interpreting these volume changes, we derive a novel method for calculating strain and normal stresses in the rebar. Our new method captures the detailed distribution of the bond stresses between rebar and concrete and consequently describes the bond behavior more accurate. It turns out that the observed bond behavior cannot be explained completely by commonly assumed material laws. This emphasizes the need for further, more extensive research combining 3D imaging, mechanical testing, and quantitative image analysis.