no code implementations • 2 Jun 2023 • Ameen Ali, Tomer Galanti, Lior Wolf
The self-attention mechanism in transformers and the message-passing mechanism in graph neural networks are repeatedly applied within deep learning architectures.
1 code implementation • NeurIPS 2023 • Ido Ben-Shaul, Ravid Shwartz-Ziv, Tomer Galanti, Shai Dekel, Yann Lecun
Self-supervised learning (SSL) is a powerful tool in machine learning, but understanding the learned representations and their underlying mechanisms remains a challenge.
no code implementations • 23 Mar 2023 • Liu Ziyin, Botao Li, Tomer Galanti, Masahito Ueda
Characterizing and understanding the stability of Stochastic Gradient Descent (SGD) remains an open problem in deep learning.
no code implementations • 28 Jan 2023 • Tomer Galanti, Mengjia Xu, Liane Galanti, Tomaso Poggio
In this paper, we investigate the Rademacher complexity of deep sparse neural networks, where each neuron receives a small number of inputs.
no code implementations • 11 Jan 2023 • Ido Ben-Shaul, Tomer Galanti, Shai Dekel
Multiplication layers are a key component in various influential neural network modules, including self-attention and hypernetwork layers.
no code implementations • 23 Dec 2022 • Tomer Galanti, András György, Marcus Hutter
We study the ability of foundation models to learn representations for classification that are transferable to new, unseen classes.
no code implementations • 12 Jun 2022 • Tomer Galanti, Zachary S. Siegel, Aparna Gupte, Tomaso Poggio
We study the bias of Stochastic Gradient Descent (SGD) to learn low-rank weight matrices when training deep neural networks.
no code implementations • 18 Feb 2022 • Tomer Galanti, Liane Galanti, Ido Ben-Shaul
Finally, we empirically show that the effective depth of a trained neural network monotonically increases when increasing the number of random labels in data.
no code implementations • ICLR 2022 • Tomer Galanti, András György, Marcus Hutter
We study the ability of foundation models to learn representations for classification that are transferable to new, unseen classes.
1 code implementation • NeurIPS 2021 • Raphael Bensadoun, Shir Gur, Tomer Galanti, Lior Wolf
Internal learning for single-image generation is a framework, where a generator is trained to produce novel images based on a single image.
1 code implementation • 8 Jun 2021 • Chenfeng Xu, Shijia Yang, Tomer Galanti, Bichen Wu, Xiangyu Yue, Bohan Zhai, Wei Zhan, Peter Vajda, Kurt Keutzer, Masayoshi Tomizuka
We discover that we can indeed use the same architecture and pretrained weights of a neural net model to understand both images and point-clouds.
no code implementations • 22 Mar 2021 • Ameen Ali, Tomer Galanti, Evgeniy Zheltonozhskiy, Chaim Baskin, Lior Wolf
We consider the problem of the extraction of semantic attributes, supervised only with classification labels.
no code implementations • 26 Apr 2020 • Yaniv Benny, Tomer Galanti, Sagie Benaim, Lior Wolf
We present two new metrics for evaluating generative models in the class-conditional image generation setting.
1 code implementation • NeurIPS 2020 • Etai Littwin, Tomer Galanti, Lior Wolf, Greg Yang
{\em Hypernetworks} are architectures that produce the weights of a task-specific {\em primary network}.
no code implementations • 23 Feb 2020 • Tomer Galanti, Ofir Nabati, Lior Wolf
In the multivariate case, where one can ensure that the complexities of the cause and effect are balanced, we propose a new adversarial training method that mimics the disentangled structure of the causal model.
1 code implementation • NeurIPS 2020 • Tomer Galanti, Lior Wolf
Besides, we show that for a structured target function, the overall number of trainable parameters in a hypernetwork is smaller by orders of magnitude than the number of trainable parameters of a standard neural network and an embedding method.
no code implementations • 28 Jan 2020 • Etai Littwin, Tomer Galanti, Lior Wolf
We derive finite width and depth corrections for the Neural Tangent Kernel (NTK) of ResNets and DenseNets.
no code implementations • 15 Jan 2020 • Lior Wolf, Tomer Galanti, Tamir Hazan
We regard explanations as a blending of the input sample and the model's output and offer a few definitions that capture various desired properties of the function that generates these explanations.
1 code implementation • ICLR 2019 • Ori Press, Tomer Galanti, Sagie Benaim, Lior Wolf
Thus, in the above example, we can create, for every person without glasses a version with the glasses observed in any face image.
no code implementations • ICLR 2019 • Lior Wolf, Sagie Benaim, Tomer Galanti
Two functions are learned: (i) a set indicator c, which is a binary classifier, and (ii) a comparator function h that given two nearby samples, predicts which sample has the higher value of the unknown function v. Loss terms are used to ensure that all training samples x are a local maxima of v, according to h and satisfy c(x)=1.
no code implementations • 25 Sep 2019 • Tomer Galanti, Ofir Nabati, Lior Wolf
Comparing the reconstruction errors of the two autoencoders, one for each variable, is shown to perform well on the accepted benchmarks of the field.
1 code implementation • ICCV 2019 • Sagie Benaim, Michael Khaitov, Tomer Galanti, Lior Wolf
We present a method for recovering the shared content between two visual domains as well as the content that is unique to each domain.
no code implementations • 23 Jul 2018 • Tomer Galanti, Sagie Benaim, Lior Wolf
The recent empirical success of unsupervised cross-domain mapping algorithms, between two domains that share common characteristics, is not well-supported by theoretical justifications.
1 code implementation • ECCV 2018 • Sagie Benaim, Tomer Galanti, Lior Wolf
While in supervised learning, the validation error is an unbiased estimator of the generalization (test) error and complexity-based generalization bounds are abundant, no such bounds exist for learning a mapping in an unsupervised way.
no code implementations • ICLR 2018 • Tomer Galanti, Lior Wolf, Sagie Benaim
We discuss the feasibility of the following learning problem: given unmatched samples from two domains and nothing else, learn a mapping between the two, which preserves semantics.
no code implementations • 5 Mar 2017 • Tomer Galanti, Lior Wolf
We consider the complementary problem in which the unlabeled samples are given post mapping, i. e., we are given the outputs of the mapping of unknown samples from the shifted domain.