OLOfer Lavi

Ofer Lavi

*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


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


Latest Publications

  • Open Access

    Estimate and Replace: A Novel Approach to Integrating Deep Neural Networks with Existing Applications

    by Guy Hadash, Einat Kermany, Boaz Carmeli, Ofer Lavi, George Kour, Alon Jacovi

    Existing applications include a huge amount of knowledge that is out of reach for deep neural networks. This paper presents a novel approach for integrating calls to existing applications into deep learning architectures. Using this approach, we estimate each application's functionality with an estimator, which is implemented as a deep neural network (DNN). The estimator is then embedded into a base network that we direct into complying with the application's interface during an end-to-end optimization process. At inference time, we replace each estimator with its existing application counterpart and let the base network solve the task by interacting with the existing application. Using this 'Estimate and Replace' method, we were able to train a DNN end-to-end with less data and outperformed a matching DNN that did not interact with the external application.

  • Open Access

    Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Tasks

    by Yossi Adi, Einat Kermany, Yonatan Belinkov, Ofer Lavi, Yoav Goldberg

    There is a lot of research interest in encoding variable length sentences into fixed length vectors, in a way that preserves the sentence meanings. Two common methods include representations based on averaging word vectors, and representations based on the hidden states of recurrent neural networks such as LSTMs. The sentence vectors are used as features for subsequent machine learning tasks or for pre-training in the context of deep learning. However, not much is known about the properties that are encoded in these sentence representations and about the language information they capture. We propose a framework that facilitates better understanding of the encoded representations. We define prediction tasks around isolated aspects of sentence structure (namely sentence length, word content, and word order), and score representations by the ability to train a classifier to solve each prediction task when using the representation as input. We demonstrate the potential contribution of the approach by analyzing different sentence representation mechanisms. The analysis sheds light on the relative strengths of different sentence embedding methods with respect to these low level prediction tasks, and on the effect of the encoded vector's dimensionality on the resulting representations.


  • Network-Induced Classification Kernels for Gene Expression Profile Analysis

    by Ofer Lavi, Gideon Dror, Ron Shamir

    Computational classification of gene expression profiles into distinct disease phenotypes has been highly successful to date. Still, robustness, accuracy, and biological interpretation of the results have been limited, and it was suggested that use of protein interaction information jointly with the expression profiles can improve the results. Here, we study three aspects of this problem. First, we show that interactions are indeed relevant by showing that co-expressed genes tend to be closer in the network of interactions. Second, we show that the improved performance of one extant method utilizing expression and interactions is not really due to the biological information in the network, while in another method this is not the case. Finally, we develop a new kernel method—called NICK—that integrates network and expression data for SVM classification, and demonstrate that overall it achieves better results than extant methods while running two orders of magnitude faster.