NSNir Shavit

Nir Shavit

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Nir Shavit has a total of 47 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • SmartEM: machine learning-guided electron microscopy

    by Yaron Meirovitch, Ishaan Singh Chandok, Core Francisco Park, Pavel Potocek, Lu Mi, Shashata Sawmya, Yicong Li, Thomas L. Athey, Vladislav Susoy, Neha Karlupia, et al.

    Connectomics provides nanometer-resolution, synapse-level maps of neural circuits to understand brain activity and behavior. However, few researchers have access to the high-throughput electron microscopes necessary to generate enough data for whole-brain or even whole-circuit reconstruction. To date, machine learning methods have been used after the collection of images by electron microscopy (EM) to accelerate and improve neuronal segmentation, synapse reconstruction and other data analysis. With the continual computational improvements in processing EM images, acquiring EM images will become the rate-limiting step in automated connectomics. Here, in order to speed up EM imaging, we integrate machine learning into real-time image acquisition in a single-beam scanning electron microscope. This SmartEM approach allows an electron microscope to perform data-aware imaging of specimens. SmartEM saves time by allocating the proper imaging time for each region of interest—first scanning all pixels rapidly and then rescanning more slowly only the small subareas where a higher quality signal is required. We demonstrate that SmartEM achieves up to an ~7-fold acceleration of image acquisition time for connectomic samples using a commercial single-beam SEM in samples from nematodes, mice and human brain. We apply this fast imaging method to reconstruct a portion of mouse cerebral cortex with an accuracy comparable to traditional electron microscopy.


  • Open Access

    A connectomics-driven analysis reveals novel characterization of border regions in mouse visual cortex

    by Neehal Tumma, Linghao Kong, Shashata Sawmya, Tony T. Wang, Nir Shavit

    Leveraging retinotopic maps to parcellate the visual cortex into its respective sub-regions has long been a canonical approach to characterizing the functional organization of visual areas in the mouse brain. However, with the advent of extensive connectomics datasets like MICrONS, we can now perform more granular analyses to better characterize the structure and function of the visual cortex. In this work, we propose a statistical framework for analyzing the MICrONS dataset, particularly the V1, RL, and AL visual areas. In addition to identifying several structural and functional differences between these regions, we focus on the borders between these regions. By comparing the V1-RL and RL-AL border regions, we show that different boundaries between visual regions are distinct in their structure and function. Additionally, we find that the V1-RL border region has greater synaptic connectivity and more synchronous neural activity than the V1 and RL regions individually. We further analyze structure and function in tandem by measuring information flow along synapses, observing that the V1-RL border appears to act as a bridge between the V1 and RL visual areas. Overall, we identify numerous measures that distinguish the V1-RL border from the larger V1-RL network, potentially motivating its characterization as a distinct region in the mouse visual cortex.

  • Open Access

    Sparsity in Deep Neural Nets (Keynote)

    by Nir N Shavit

    Our brain executes very sparse computation, allowing for great speed and energy savings. Deep neural networks can also be made to exhibit high levels of sparsity without significant accuracy loss. As their size grows, it is becoming imperative that we use sparsity to improve their efficiency. This is a challenging task because the memory systems and SIMD operations that dominate todays CPUs and GPUs do not lend themselves easily to the irregular data patterns sparsity introduces. This talk will survey the role of sparsity in neural network computation, and the parallel algorithms and hardware features that nevertheless allow us to make effective use of it.

  • Open Access

    Cliff-Learning

    by Tony T. Wang, Igor Zablotchi, Nir Shavit, Jonathan S. Rosenfeld

    We study the data-scaling of transfer learning from foundation models in the low-downstream-data regime. We observe an intriguing phenomenon which we call cliff-learning. Cliff-learning refers to regions of data-scaling laws where performance improves at a faster than power law rate (i.e. regions of concavity on a log-log scaling plot). We conduct an in-depth investigation of foundation-model cliff-learning and study toy models of the phenomenon. We observe that the degree of cliff-learning reflects the degree of compatibility between the priors of a learning algorithm and the task being learned.

  • Open Access

    A Complexity-Based Hierarchy for Multiprocessor Synchronization

    by Faith Ellen, Rati Gelashvili, Nir Shavit, Leqi Zhu

    For many years, Herlihy's elegant computability based Consensus Hierarchy has been our best explanation of the relative power of various types of multiprocessor synchronization objects when used in deterministic algorithms. However, key to this hierarchy is treating synchronization instructions as distinct objects, an approach that is far from the real-world, where multiprocessor programs apply synchronization instructions to collections of arbitrary memory locations. We were surprised to realize that, when considering instructions applied to memory locations, the computability based hierarchy collapses. This leaves open the question of how to better capture the power of various synchronization instructions. In this paper, we provide an approach to answering this question. We present a hierarchy of synchronization instructions, classified by their space complexity in solving obstruction-free consensus. Our hierarchy provides a classification of combinations of known instructions that seems to fit with our intuition of how useful some are in practice, while questioning the effectiveness of others. We prove an essentially tight characterization of the power of buffered read and write instructions.Interestingly, we show a similar result for multi-location atomic assignments.

  • Open Access

    Inherent limitations of hybrid transactional memory

    by Dan Alistarh, Justin Kopinsky, Petr Kuznetsov, Srivatsan Ravi, Nir Shavit


  • Open Access

    A Complexity-Based Hierarchy for Multiprocessor Synchronization

    by Faith Ellen, Rati Gelashvili, Nir Shavit, Leqi Zhu

    For many years, Herlihy's elegant computability based Consensus Hierarchy has been our best explanation of the relative power of various types of multiprocessor synchronization objects when used in deterministic algorithms. However, key to this hierarchy is treating these instructions as distinct objects, an approach that is far from the real-world, where multiprocessor programs apply synchronization instructions to collections of arbitrary memory locations. We were surprised to realize that, when considering instructions applied to memory locations, the computability based hierarchy collapses. This leaves open the question of how to better captures the power of various synchronization instructions. In this paper, we provide an approach to answering this question. We present a hierarchy of synchronization instructions, classified by their space complexity in solving obstruction-free consensus. Our hierarchy provides a classification of combinations of known instructions that seems to fit with our intuition of how useful some are in practice, while questioning the effectiveness of others. We prove an essentially tight characterization of the power of buffered read and write instructions. Interestingly, we show a similar result for multi-location atomic assignments.


  • Transactional Memory

    by Nir Shavit, Alexander Matveev

  • Amalgamated Lock-Elision

    by Yehuda Afek, Alexander Matveev, Oscar R. Moll, Nir Shavit