@article{17909, abstract = {Atomically precise clusters can be used to create single-electron devices wherein a single redox-active cluster is connected to two macroscopic electrodes via anchoring ligands. Unlike single-electron devices comprising nanocrystals, these cluster-based devices can be fabricated with atomic precision. This affords an unprecedented level of control over the device properties. Herein, we design a series of cobalt chalcogenide clusters with varying ligand geometries and core nuclearities to control their current–voltage (I–V) characteristics in a scanning tunneling microscope-based break junction (STM-BJ) device. First, the device geometry is modified by precisely positioning junction-anchoring ligands on the surface of the cluster. We show that the I–V characteristics are independent of ligand placement, confirming a sequential, single-electron tunneling mechanism. Next, we chemically fuse two clusters to realize a larger cluster dimer that behaves as a single electronic unit, possessing a smaller reorganization energy and more accessible redox states than the monomeric analogues. As a result, dimer-based devices exhibit significantly higher currents and can even be pushed to current saturation at high bias. Owing to these controllable properties, single-cluster junctions serve as an excellent platform for exploring incoherent charge transport processes at the nanoscale. With this understanding, as well as properties such as nonlinear I–V characteristics and rectification, these molecular clusters may function as conductive inorganic nodes in new devices and materials.}, author = {Gunasekaran, Suman and Reed, Douglas A. and Paley, Daniel W. and Bartholomew, Amymarie K. and Venkataraman, Latha and Steigerwald, Michael L. and Roy, Xavier and Nuckolls, Colin}, issn = {1520-5126}, journal = {Journal of the American Chemical Society}, number = {35}, pages = {14924--14932}, publisher = {American Chemical Society}, title = {{Single-electron currents in designer single-cluster devices}}, doi = {10.1021/jacs.0c04970}, volume = {142}, year = {2020}, } @article{17910, abstract = {The scanning tunneling microscope-based break junction (STM-BJ) is used widely to create and characterize single metal-molecule-metal junctions. In this technique, conductance is continuously recorded as a metal point contact is broken in a solution of molecules. Conductance plateaus are seen when stable molecular junctions are formed. Typically, thousands of junctions are created and measured, yielding thousands of distinct conductance versus extension traces. However, such traces are rarely analyzed individually to recognize the types of junctions formed. Here, we present a deep learning-based method to identify molecular junctions and show that it performs better than several commonly used and recently reported techniques. We demonstrate molecular junction identification from mixed solution measurements with accuracies as high as 97%. We also apply this model to an in situ electric field-driven isomerization reaction of a [3]cumulene to follow the reaction over time. Furthermore, we demonstrate that our model can remain accurate even when a key parameter, the average junction conductance, is eliminated from the analysis, showing that our model goes beyond conventional analysis in existing methods.}, author = {Fu, Tianren and Zang, Yaping and Zou, Qi and Nuckolls, Colin and Venkataraman, Latha}, issn = {1530-6992}, journal = {Nano Letters}, number = {5}, pages = {3320--3325}, publisher = {American Chemical Society}, title = {{Using deep learning to identify molecular junction characteristics}}, doi = {10.1021/acs.nanolett.0c00198}, volume = {20}, year = {2020}, } @article{17911, abstract = {Single-molecule break junction measurements deliver a huge number of conductance vs. electrode separation traces. During such measurements, the target molecules may bind to the electrodes in different geometries, and the evolution and rupture of the single-molecule junction may also follow distinct trajectories. The unraveling of the various typical trace classes is a prerequisite to the proper physical interpretation of the data. Here we exploit the efficient feature recognition properties of neural networks to automatically find the relevant trace classes. To eliminate the need for manually labeled training data we apply a combined method, which automatically selects training traces according to the extreme values of principal component projections or some auxiliary measured quantities. Then the network captures the features of these characteristic traces and generalizes its inference to the entire dataset. The use of a simple neural network structure also enables a direct insight into the decision-making mechanism. We demonstrate that this combined machine learning method is efficient in the unsupervised recognition of unobvious, but highly relevant trace classes within low and room temperature gold–4,4′ bipyridine–gold single-molecule break junction data.}, author = {Magyarkuti, András and Balogh, Nóra and Balogh, Zoltán and Venkataraman, Latha and Halbritter, András}, issn = {2040-3372}, journal = {Nanoscale}, number = {15}, pages = {8355--8363}, publisher = {Royal Society of Chemistry}, title = {{Unsupervised feature recognition in single-molecule break junction data}}, doi = {10.1039/d0nr00467g}, volume = {12}, year = {2020}, } @article{17912, abstract = {Aryl halides are ubiquitous functional groups in organic chemistry, yet despite their obvious appeal as surface-binding linkers and as precursors for controlled graphene nanoribbon synthesis, they have seldom been used as such in molecular electronics. The confusion regarding the bonding of aryl iodides to Au electrodes is a case in point, with ambiguous reports of both dative Au–I and covalent Au–C contacts. Here we form single-molecule junctions with a series of oligophenylene molecular wires terminated asymmetrically with iodine and thiomethyl to show that the dative Au–I contact has a lower conductance than the covalent Au–C interaction, which we propose occurs via an in situ oxidative addition reaction at the Au surface. Furthermore, we confirm the formation of the Au–C bond by measuring an analogous series of molecules prepared ex situ with the complex AuI(PPh3) in place of the iodide. Density functional theory-based transport calculations support our experimental observations that Au–C linkages have higher conductance than Au–I linkages. Finally, we demonstrate selective promotion of the Au–C bond formation by controlling the bias applied across the junction. In addition to establishing the different binding modes of aryl iodides, our results chart a path to actively controlling oxidative addition on an Au surface using an applied bias.}, author = {Starr, Rachel L. and Fu, Tianren and Doud, Evan A. and Stone, Ilana and Roy, Xavier and Venkataraman, Latha}, issn = {1520-5126}, journal = {Journal of the American Chemical Society}, number = {15}, pages = {7128--7133}, publisher = {American Chemical Society}, title = {{Gold–carbon contacts from oxidative addition of aryl iodides}}, doi = {10.1021/jacs.0c01466}, volume = {142}, year = {2020}, } @article{17913, abstract = {Electron transport across a molecular junction is characterized by an energy-dependent transmission function. The transmission function accounts for electrons tunneling through multiple molecular orbitals (MOs) with different phases, which gives rise to quantum interference (QI) effects. Because the transmission function comprises both interfering and noninterfering effects, individual interferences between MOs cannot be deduced from the transmission function directly. Herein, we demonstrate how the transmission function can be deconstructed into its constituent interfering and noninterfering contributions for any model molecular junction. These contributions are arranged in a matrix and displayed pictorially as a QI map, which allows one to easily identify individual QI effects. Importantly, we show that exponential conductance decay with increasing oligomer length is primarily due to an increase in destructive QI. With an ability to “see” QI effects using the QI map, we find that QI is vital to all molecular-scale electron transport.}, author = {Gunasekaran, Suman and Greenwald, Julia E. and Venkataraman, Latha}, issn = {1530-6992}, journal = {Nano Letters}, number = {4}, pages = {2843--2848}, publisher = {American Chemical Society}, title = {{Visualizing quantum interference in molecular junctions}}, doi = {10.1021/acs.nanolett.0c00605}, volume = {20}, year = {2020}, } @article{17914, abstract = {Polyacetylene molecular wires have attracted a long-standing interest for the past 40 years. From a fundamental perspective, there are two main reasons for the interest. First, polyacetylenes are a prime realization of a one-dimensional topological insulator. Second, long molecules support freely propagating topological domain-wall states, so-called “solitons,” which provide an early paradigm for spin-charge separation. Because of recent experimental developments, individual polyacetylene chains can now be synthesized on substrates. Motivated by this breakthrough, we here propose a novel way for chemically supported soliton design in these systems. We demonstrate how to control the soliton position and how to read it out via external means. Also, we show how extra soliton–antisoliton pairs arise when applying a moderate static electric field. We thus make a step toward functionality of electronic devices based on soliton manipulation, that is, “solitonics”.}, author = {Hernangómez-Pérez, Daniel and Gunasekaran, Suman and Venkataraman, Latha and Evers, Ferdinand}, issn = {1530-6992}, journal = {Nano Letters}, number = {4}, pages = {2615--2619}, publisher = {American Chemical Society}, title = {{Solitonics with polyacetylenes}}, doi = {10.1021/acs.nanolett.0c00136}, volume = {20}, year = {2020}, } @article{18194, abstract = {Realizing strongly correlated topological phases of ultracold gases is a central goal for ongoing experiments. While fractional quantum Hall states could soon be implemented in small atomic ensembles, detecting their signatures in few-particle settings remains a fundamental challenge. In this work, we numerically analyze the center-of-mass Hall drift of a small ensemble of hardcore bosons, initially prepared in the ground state of the Harper-Hofstadter-Hubbard model in a box potential. By monitoring the Hall drift upon release, for a wide range of magnetic flux values, we identify an emergent Hall plateau compatible with a fractional Chern insulator state: The extracted Hall conductivity approaches a fractional value determined by the many-body Chern number, while the width of the plateau agrees with the spectral and topological properties of the prepared ground state. Besides, a direct application of Streda's formula indicates that such Hall plateaus can also be directly obtained from static density-profile measurements. Our calculations suggest that fractional Chern insulators can be detected in cold-atom experiments, using available detection methods.}, author = {Repellin, C. and Leonard, Julian and Goldman, N.}, issn = {2469-9934}, journal = {Physical Review A}, number = {6}, publisher = {American Physical Society}, title = {{Fractional Chern insulators of few bosons in a box: Hall plateaus from center-of-mass drifts and density profiles}}, doi = {10.1103/physreva.102.063316}, volume = {102}, year = {2020}, } @article{18228, abstract = {We introduce two constructions in geometric deep learning for 1) transporting orientation-dependent convolutional filters over a manifold in a continuous way and thereby defining a convolution operator that naturally incorporates the rotational effect of holonomy; and 2) allowing efficient evaluation of manifold convolution layers by sampling manifold valued random variables that center around a weighted diffusion mean. Both methods are inspired by stochastics on manifolds and geometric statistics, and provide examples of how stochastic methods – here horizontal frame bundle flows and non-linear bridge sampling schemes, can be used in geometric deep learning. We outline the theoretical foundation of the two methods, discuss their relation to Euclidean deep networks and existing methodology in geometric deep learning, and establish important properties of the proposed constructions.}, author = {Sommer, Stefan and Bronstein, Alexander}, issn = {1939-3539}, journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, number = {2}, pages = {811--822}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{Horizontal flows and manifold stochastics in geometric deep learning}}, doi = {10.1109/tpami.2020.2994507}, volume = {44}, year = {2020}, } @article{18245, abstract = {Intel® RealSense™ SR300 is a depth camera capable of providing a VGA-size depth map at 60 fps and 0.125mm depth resolution. In addition, it outputs an infrared VGA-resolution image and a 1080p color texture image at 30 fps. SR300 form-factor enables it to be integrated into small consumer products and as a front facing camera in laptops and Ultrabooks™. The SR300 depth camera is based on a coded-light technology where triangulation between projected patterns and images captured by a dedicated sensor is used to produce the depth map. Each projected line is coded by a special temporal optical code, that enables a dense depth map reconstruction from its reflection. The solid mechanical assembly of the camera allows it to stay calibrated throughout temperature and pressure changes, drops, and hits. In addition, active dynamic control maintains a calibrated depth output. An extended API LibRS released with the camera allows developers to integrate the camera in various applications. Algorithms for 3D scanning, facial analysis, hand gesture recognition, and tracking are within reach for applications using the SR300. In this paper, we describe the underlying technology, hardware, and algorithms of the SR300, as well as its calibration procedure, and outline some use cases. We believe that this paper will provide a full case study of a mass-produced depth sensing product and technology.}, author = {Zabatani, Aviad and Surazhsky, Vitaly and Sperling, Erez and Moshe, Sagi Ben and Menashe, Ohad and Silver, David H. and Karni, Zachi and Bronstein, Alexander and Bronstein, Michael K and Kimmel, Ron}, issn = {1939-3539}, journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence}, number = {10}, pages = {2333--2345}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{Intel® RealSense™ SR300 coded light depth camera}}, doi = {10.1109/tpami.2019.2915841}, volume = {42}, year = {2020}, } @article{18246, abstract = {Adversarial attacks have demonstrated remarkable efficacy in altering the output of a learning model by applying a minimal perturbation to the input data. While increasing attention has been placed on the image domain, however, the study of adversarial perturbations for geometric data has been notably lagging behind. In this paper, we show that effective adversarial attacks can be concocted for surfaces embedded in 3D, under weak smoothness assumptions on the perceptibility of the attack. We address the case of deformable 3D shapes in particular, and introduce a general model that is not tailored to any specific surface representation, nor does it assume access to a parametric description of the 3D object. In this context, we consider targeted and untargeted variants of the attack, demonstrating compelling results in either case. We further show how discovering adversarial examples, and then using them for adversarial training, leads to an increase in both robustness and accuracy. Our findings are confirmed empirically over multiple datasets spanning different semantic classes and deformations.}, author = {Mariani, G. and Cosmo, L. and Bronstein, Alexander and Rodolà, E.}, issn = {1467-8659}, journal = {Computer Graphics Forum}, number = {5}, pages = {253--264}, publisher = {Wiley}, title = {{Generating adversarial surfaces via band‐limited perturbations}}, doi = {10.1111/cgf.14083}, volume = {39}, year = {2020}, } @inproceedings{18247, abstract = {Convolutional neural networks (CNNs) achieve state-of-the-art accuracy in a variety of tasks in computer vision and beyond. One of the major obstacles hindering the ubiquitous use of CNNs for inference on low-power edge devices is their high computational complexity and memory bandwidth requirements. The latter often dominates the energy footprint on modern hardware. In this paper, we introduce a lossy transform coding approach, inspired by image and video compression, designed to reduce the memory bandwidth due to the storage of intermediate activation calculation results. Our method does not require fine-tuning the network weights and halves the data transfer volumes to the main memory by compressing feature maps, which are highly correlated, with variable length coding. Our method outperform previous approach in term of the number of bits per value with minor accuracy degradation on ResNet-34 and MobileNetV2. We analyze the performance of our approach on a variety of CNN architectures and demonstrate that FPGA implementation of ResNet-18 with our approach results in a reduction of around 40% in the memory energy footprint, compared to quantized network, with negligible impact on accuracy. When allowing accuracy degradation of up to 2%, the reduction of 60% is achieved. A reference implementation accompanies the paper.}, author = {Chmiel, Brian and Baskin, Chaim and Zheltonozhskii, Evgenii and Banner, Ron and Yermolin, Yevgeny and Karbachevsky, Alex and Bronstein, Alexander and Mendelson, Avi}, booktitle = {2020 International Joint Conference on Neural Networks (IJCNN)}, isbn = {9781728169279}, issn = {2161-4407}, location = {Glasgow, United Kingdom}, publisher = {IEEE}, title = {{Feature map transform coding for energy-efficient CNN inference}}, doi = {10.1109/ijcnn48605.2020.9206968}, year = {2020}, } @inproceedings{18248, abstract = {Learning an object detection or retrieval system requires a large data set with manual annotations. Such data are expensive and time-consuming to create and therefore difficult to obtain on a large scale. In this work, we propose using the natural correlation in narrations and the visual presence of objects in video to learn an object detector and retriever without any manual labeling involved. We pose the problem as weakly supervised learning with noisy labels, and propose a novel object detection and retrieval paradigm under these constraints. We handle the background rejection by using contrastive samples and confront the high level of label noise with a new clustering score. Our evaluation is based on a set of ten objects with manual ground truth annotation in almost 5000 frames extracted from instructional videos from the web. We demonstrate superior results compared to state-of-the-art weakly- supervised approaches and report a strongly-labeled upper bound as well. While the focus of the paper is object detection and retrieval, the proposed methodology can be applied to a broader range of noisy weakly-supervised problems.}, author = {Amrani, Elad and Ben-Ari, Rami and Shapira, Inbar and Hakim, Tal and Bronstein, Alexander}, booktitle = {2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)}, isbn = {9781728193618}, issn = {2160-7516}, location = {Seattle, WA, United States}, publisher = {IEEE}, title = {{Self-supervised object detection and retrieval using unlabeled videos}}, doi = {10.1109/cvprw50498.2020.00485}, year = {2020}, } @inproceedings{18249, abstract = {Magnetic Resonance Imaging (MRI) is considered today the golden-standard modality for soft tissues. The long acquisition times, however, make it more prone to motion artifacts as well as contribute to the relative high costs of this examination. Over the years, multiple studies concentrated on designing reduced measurement schemes and image reconstruction schemes for MRI, however these problems have been so far addressed separately. On the other hand, recent works in optical computational imaging have demonstrated growing success of simultaneous learning-based design of the acquisition and reconstruction schemes manifesting significant improvement in the reconstruction quality with a constrained time budget. Inspired by these successes, in this work, we propose to learn accelerated MR acquisition schemes (in the form of Cartesian trajectories) jointly with the image reconstruction operator. To this end, we propose an algorithm for training the combined acquisition-reconstruction pipeline end-to-end in a differentiable way. We demonstrate the significance of using the learned Cartesian trajectories at different speed up rates. Code available at https://github.com/tomer196/fastMRI-Cartesian.}, author = {Weiss, Tomer and Vedula, Sanketh and Senouf, Ortal and Michailovich, Oleg and Zibulevsky, Michael and Bronstein, Alexander}, booktitle = {ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)}, isbn = {9781509066322}, issn = {2379-190X}, location = {Barcelona, Spain}, publisher = {IEEE}, title = {{Joint learning of cartesian undersampling and reconstruction for accelerated MRI}}, doi = {10.1109/icassp40776.2020.9054542}, year = {2020}, } @article{18250, abstract = {Many shape analysis methods treat the geometry of an object as a metric space that can be captured by the Laplace-Beltrami operator. In this paper, we propose to adapt the classical Hamiltonian operator from quantum mechanics to the field of shape analysis. To this end, we study the addition of a potential function to the Laplacian as a generator for dual spaces in which shape processing is performed. We present general optimization approaches for solving variational problems involving the basis defined by the Hamiltonian using perturbation theory for its eigenvectors. The suggested operator is shown to produce better functional spaces to operate with, as demonstrated on different shape analysis tasks.}, author = {Choukroun, Yoni and Shtern, Alon and Bronstein, Alexander and Kimmel, Ron}, issn = {2160-9306}, journal = {IEEE Transactions on Visualization and Computer Graphics}, number = {2}, pages = {1320--1331}, publisher = {Institute of Electrical and Electronics Engineers}, title = {{Hamiltonian operator for spectral shape analysis}}, doi = {10.1109/tvcg.2018.2867513}, volume = {26}, year = {2020}, } @inproceedings{18251, abstract = {Magnetic Resonance Imaging (MRI) has long been considered to be among the gold standards of today’s diagnostic imaging. The most significant drawback of MRI is long acquisition times, prohibiting its use in standard practice for some applications. Compressed sensing (CS) proposes to subsample the k-space (the Fourier domain dual to the physical space of spatial coordinates) leading to significantly accelerated acquisition. However, the benefit of compressed sensing has not been fully exploited; most of the sampling densities obtained through CS do not produce a trajectory that obeys the stringent constraints of the MRI machine imposed in practice. Inspired by recent success of deep learning-based approaches for image reconstruction and ideas from computational imaging on learning-based design of imaging systems, we introduce 3D FLAT, a novel protocol for data-driven design of 3D non-Cartesian accelerated trajectories in MRI. Our proposal leverages the entire 3D k-space to simultaneously learn a physically feasible acquisition trajectory with a reconstruction method. Experimental results, performed as a proof-of-concept, suggest that 3D FLAT achieves higher image quality for a given readout time compared to standard trajectories such as radial, stack-of-stars, or 2D learned trajectories (trajectories that evolve only in the 2D plane while fully sampling along the third dimension). Furthermore, we demonstrate evidence supporting the significant benefit of performing MRI acquisitions using non-Cartesian 3D trajectories over 2D non-Cartesian trajectories acquired slice-wise.}, author = {Alush-Aben, Jonathan and Ackerman-Schraier, Linor and Weiss, Tomer and Vedula, Sanketh and Senouf, Ortal and Bronstein, Alexander}, booktitle = {International Workshop on Machine Learning for Medical Image Reconstruction}, isbn = {9783030615970}, issn = {1611-3349}, location = {Lima, Peru}, pages = {3 -- 16}, publisher = {Springer Nature}, title = {{3D FLAT: Feasible learned acquisition trajectories for accelerated MRI}}, doi = {10.1007/978-3-030-61598-7_1}, volume = {12450}, year = {2020}, } @article{18252, abstract = {Vector fields arise in a variety of quantity measure and visualization techniques, such as fluid flow imaging, motion estimation, deformation measures, and color imaging, leading to a better understanding of physical phenomena. Recent progress in vector field imaging technologies has emphasized the need for efficient noise removal and reconstruction algorithms. A key ingredient in the successful extraction of signals from noisy measurements is prior information, which can often be represented as a parameterized model. In this work, we extend the overparameterization variational framework in order to perform model-based reconstruction of vector fields. The overparameterization methodology combines local modeling of the data with global model parameter regularization. By considering the vector field as a linear combination of basis vector fields and appropriate scale and rotation coefficients, we can reduce the denoising problem to a simpler form of coefficient recovery. We introduce two versions of the overparameterization framework: a total variation-based method and a sparsity-based method, which relies on the cosparse analysis model. We demonstrate the efficiency of the proposed frameworks for two- and three-dimensional vector fields with linear and quadratic overparameterization models.}, author = {Rotker, Keren and Bashat, Dafna Ben and Bronstein, Alexander}, issn = {1936-4954}, journal = {SIAM Journal on Imaging Sciences}, number = {3}, pages = {1386--1414}, publisher = {Society for Industrial & Applied Mathematics}, title = {{Overparameterized models for vector fields}}, doi = {10.1137/19m1280697}, volume = {13}, year = {2020}, } @article{18253, abstract = {PCNA, the ring that encircles DNA maintaining the processivity of DNA polymerases, is modified by ubiquitin and SUMO. Whereas ubiquitin is required for bypassing lesions through the DNA damage tolerance (DDT) pathways, we show here that SUMOylation represses another pathway, salvage recombination. The Srs2 helicase is recruited to SUMOylated PCNA and prevents the salvage pathway from acting. The pathway can be induced by overexpressing the PCNA unloader Elg1, or the homologous recombination protein Rad52. Our results underscore the role of PCNA modifications in controlling the various bypass and DNA repair mechanisms.}, author = {Arbel, Matan and Bronstein, Alexander and Sau, Soumitra and Liefshitz, Batia and Kupiec, Martin}, issn = {2150-7511}, journal = {mBio}, number = {3}, publisher = {American Society for Microbiology}, title = {{Access to PCNA by Srs2 and Elg1 controls the choice between alternative repair pathways in Saccharomyces cerevisiae}}, doi = {10.1128/mbio.00705-20}, volume = {11}, year = {2020}, } @inproceedings{18255, abstract = {Learning an object detection or retrieval system requires a large data set with manual annotations. Such data sets are expensive and time consuming to create and therefore difficult to obtain on a large scale. In this work, we propose to exploit the natural correlation in narrations and the visual presence of objects in video, to learn an object detector and retrieval without any manual labeling involved. We pose the problem as weakly supervised learning with noisy labels, and propose a novel object detection paradigm under these constraints. We handle the background rejection by using contrastive samples and confront the high level of label noise with a new clustering score. Our evaluation is based on a set of 11 manually annotated objects in over 5000 frames. We show comparison to a weakly-supervised approach as baseline and provide a strongly labeled upper bound.}, author = {Amrani, Elad and Ben-Ari, Rami and Hakim, Tal and Bronstein, Alexander}, booktitle = {2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW)}, isbn = {9781728150246}, issn = {2473-9944}, location = {Seoul, Korea (South)}, publisher = {IEEE}, title = {{Learning to detect and retrieve objects from unlabeled videos}}, doi = {10.1109/iccvw.2019.00567}, year = {2020}, } @inproceedings{18258, abstract = {Distance metric learning (DML) has been successfully applied to object classification, both in the standard regime of rich training data and in the few-shot scenario, where each category is represented by only a few examples. In this work, we propose a new method for DML that simultaneously learns the backbone network parameters, the embedding space, and the multi-modal distribution of each of the training categories in that space, in a single end-to-end training process. Our approach outperforms state-of-the-art methods for DML-based object classification on a variety of standard fine-grained datasets. Furthermore, we demonstrate the effectiveness of our approach on the problem of few-shot object detection, by incorporating the proposed DML architecture as a classification head into a standard object detection model. We achieve the best results on the ImageNet-LOC dataset compared to strong baselines, when only a few training examples are available. We also offer the community a new episodic benchmark based on the ImageNet dataset for the few-shot object detection task.}, author = {Karlinsky, Leonid and Shtok, Joseph and Harary, Sivan and Schwartz, Eli and Aides, Amit and Feris, Rogerio and Giryes, Raja and Bronstein, Alexander}, booktitle = {2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, isbn = {9781728132945}, issn = {2575-7075}, location = {Long Beach, CA, United States}, publisher = {IEEE}, title = {{Repmet: Representative-based metric learning for classification and few-shot object detection}}, doi = {10.1109/cvpr.2019.00534}, year = {2020}, } @inproceedings{18259, abstract = {Example synthesis is one of the leading methods to tackle the problem of few-shot learning, where only a small number of samples per class are available. However, current synthesis approaches only address the scenario of a single category label per image. In this work, we propose a novel technique for synthesizing samples with multiple labels for the (yet unhandled) multi-label few-shot classification scenario. We propose to combine pairs of given examples in feature space, so that the resulting synthesized feature vectors will correspond to examples whose label sets are obtained through certain set operations on the label sets of the corresponding input pairs. Thus, our method is capable of producing a sample containing the intersection, union or set-difference of labels present in two input samples. As we show, these set operations generalize to labels unseen during training. This enables performing augmentation on examples of novel categories, thus, facilitating multi-label few-shot classifier learning. We conduct numerous experiments showing promising results for the label-set manipulation capabilities of the proposed approach, both directly (using the classification and retrieval metrics), and in the context of performing data augmentation for multi-label few-shot learning. We propose a benchmark for this new and challenging task and show that our method compares favorably to all the common baselines.}, author = {Alfassy, Amit and Karlinsky, Leonid and Aides, Amit and Shtok, Joseph and Harary, Sivan and Feris, Rogerio and Giryes, Raja and Bronstein, Alexander}, booktitle = {2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)}, isbn = {9781728132945}, issn = {2575-7075}, location = {Long Beach, CA, United States}, publisher = {IEEE}, title = {{Laso: Label-set operations networks for multi-label few-shot learning}}, doi = {10.1109/cvpr.2019.00671}, year = {2020}, }