• Publicationsclose
• Research softwareclose
• 2019-2023close
• Preprint close
• Article close
• IL close
• AE close
• arXiv.org e-Print Archive close

This paper describes a novel approach to systematically improve information interactions based solely on its wording. Following an interdisciplinary literature review, we recognized three key attributes of words that drive user engagement: (1) Novelty (2) Familiarity (3) Emotionality. Based on these attributes, we developed a model to systematically improve a given content using computational linguistics, natural language processing (NLP) and text analysis (word frequency, sentiment analysis and lexical substitution). We conducted a pilot study (n=216) in which the model was used to formalize evaluation and optimization of academic titles. A between-group design (A/B testing) was used to compare responses to the original and modified (treatment) titles. Data was collected for selection and evaluation (User Engagement Scale). The pilot results suggest that user engagement with digital information is fostered by, and perhaps dependent upon, the wording being used. They also provide empirical support that engaging content can be systematically evaluated and produced. The preliminary results show that the modified (treatment) titles had significantly higher scores for information use and user engagement (selection and evaluation). We propose that computational linguistics is a useful approach for optimizing information interactions. The empirically based insights can inform the development of digital content strategies, thereby improving the success of information interactions.elop more sophisticated interaction measures. Comment: Dvir, N., & Gafni, R. (2019). Systematic improvement of user engagement with academic titles using computational linguistics. Proceedings of the Informing Science and Information Technology Education Conference, Jerusalem, Israel, pp. 501-512 Santa Rosa, CA: Informing Science Institute. https://doi.org/10.28945/4338

Policy evaluation in reinforcement learning is often conducted using two-timescale stochastic approximation, which results in various gradient temporal difference methods such as GTD(0), GTD2, and TDC. Here, we provide convergence rate bounds for this suite of algorithms. Algorithms such as these have two iterates, $\theta_n$ and $w_n,$ which are updated using two distinct stepsize sequences, $\alpha_n$ and $\beta_n,$ respectively. Assuming $\alpha_n = n^{-\alpha}$ and $\beta_n = n^{-\beta}$ with $1 > \alpha > \beta > 0,$ we show that, with high probability, the two iterates converge to their respective solutions $\theta^*$ and $w^*$ at rates given by $\|\theta_n - \theta^*\| = \tilde{O}( n^{-\alpha/2})$ and $\|w_n - w^*\| = \tilde{O}(n^{-\beta/2});$ here, $\tilde{O}$ hides logarithmic terms. Via comparable lower bounds, we show that these bounds are, in fact, tight. To the best of our knowledge, ours is the first finite-time analysis which achieves these rates. While it was known that the two timescale components decouple asymptotically, our results depict this phenomenon more explicitly by showing that it in fact happens from some finite time onwards. Lastly, compared to existing works, our result applies to a broader family of stepsizes, including non-square summable ones.

For decades, gamma-ray bursts (GRBs) have been broadly divided into `long'- and `short'-duration bursts, lasting more or less than 2s, respectively. However, this dichotomy does not map perfectly to the two progenitor channels that are known to produce GRBs -- the merger of compact objects (merger-GRBs) or the collapse of massive stars (collapsar-GRBs). In particular, the merger-GRBs population may also include bursts with a short, hard $\lesssim$2s spike and subsequent longer, softer extended emission (EE). The recent discovery of a kilonova -- the radioactive glow of heavy elements made in neutron star mergers -- in the 50s-duration GRB 211211A further demonstrates that mergers can drive long, complex GRBs that mimic the collapsar population. Here we present a detailed temporal and spectral analysis of the high-energy emission of GRB 211211A. We demonstrate that the emission has a purely synchrotron origin, with both the peak and cooling frequencies moving through the $\gamma$-ray band down to the X-rays, and that the rapidly-evolving spectrum drives the EE signature at late times. The identification of such spectral evolution in a merger-GRB opens avenues for diagnostics of the progenitor type. Comment: Author's final submitted version. 6 figures, 5 tables. The Supplementary Information .tex file is included

We study perturbative renormalization of the composite operators in the $T\bar T$-deformed two-dimensional free field theories. The pattern of renormalization for the stress-energy tensor is different in the massive and massless cases. While in the latter case the canonical stress tensor is not renormalized up to high order in the perturbative expansion, in the massive theory there are induced counterterms at linear order. For a massless theory our results match the general formula derived recently in [1]. Comment: v2: references added, version accepted to JHEP

Person re-identification has achieved great progress with deep convolutional neural networks. However, most previous methods focus on learning individual appearance feature embedding, and it is hard for the models to handle difficult situations with different illumination, large pose variance and occlusion. In this work, we take a step further and consider employing context information for person search. For a probe-gallery pair, we first propose a contextual instance expansion module, which employs a relative attention module to search and filter useful context information in the scene. We also build a graph learning framework to effectively employ context pairs to update target similarity. These two modules are built on top of a joint detection and instance feature learning framework, which improves the discriminativeness of the learned features. The proposed framework achieves state-of-the-art performance on two widely used person search datasets. To appear in CVPR 2019

Recent work on adversarial learning has focused mainly on neural networks and domains where those networks excel, such as computer vision, or audio processing. The data in these domains is typically homogeneous, whereas heterogeneous tabular datasets domains remain underexplored despite their prevalence. When searching for adversarial patterns within heterogeneous input spaces, an attacker must simultaneously preserve the complex domain-specific validity rules of the data, as well as the adversarial nature of the identified samples. As such, applying adversarial manipulations to heterogeneous datasets has proved to be a challenging task, and no generic attack method was suggested thus far. We, however, argue that machine learning models trained on heterogeneous tabular data are as susceptible to adversarial manipulations as those trained on continuous or homogeneous data such as images. To support our claim, we introduce a generic optimization framework for identifying adversarial perturbations in heterogeneous input spaces. We define distribution-aware constraints for preserving the consistency of the adversarial examples and incorporate them by embedding the heterogeneous input into a continuous latent space. Due to the nature of the underlying datasets We focus on $\ell_0$ perturbations, and demonstrate their applicability in real life. We demonstrate the effectiveness of our approach using three datasets from different content domains. Our results demonstrate that despite the constraints imposed on input validity in heterogeneous datasets, machine learning models trained using such data are still equally susceptible to adversarial examples.

We propose a data-driven approach for deep convolutional neural network compression that achieves high accuracy with high throughput and low memory requirements. Current network compression methods either find a low-rank factorization of the features that requires more memory, or select only a subset of features by pruning entire filter channels. We propose the Cascaded Projection (CaP) compression method that projects the output and input filter channels of successive layers to a unified low dimensional space based on a low-rank projection. We optimize the projection to minimize classification loss and the difference between the next layer's features in the compressed and uncompressed networks. To solve this non-convex optimization problem we propose a new optimization method of a proxy matrix using backpropagation and Stochastic Gradient Descent (SGD) with geometric constraints. Our cascaded projection approach leads to improvements in all critical areas of network compression: high accuracy, low memory consumption, low parameter count and high processing speed. The proposed CaP method demonstrates state-of-the-art results compressing VGG16 and ResNet networks with over 4x reduction in the number of computations and excellent performance in top-5 accuracy on the ImageNet dataset before and after fine-tuning.

A hierarchical Bayesian model is applied to the Cosmicflows-3 catalog of galaxy distances in order to derive the peculiar velocity field and distribution of matter within $z \sim 0.054$. The model assumes the $\Lambda$CDM model within the linear regime and includes the fit of the galaxy distances together with the underlying density field. By forward modeling the data, the method is able to mitigate biases inherent to peculiar velocity analyses, such as the Homogeneous Malmquist bias or the log-normal distribution of peculiar velocities. The statistical uncertainty on the recovered velocity field is about 150 km/s depending on the location, and we study systematics coming from the selection function and calibration of distance indicators. The resulting velocity field and related density fields recover the cosmography of the Local Universe which is presented in an unprecedented volume of universe 10 times larger than previously reached. This methodology open the doors to reconstruction of initial conditions for larger and more accurate constrained cosmological simulations. This work is also preparatory to larger peculiar velocity datasets coming from Wallaby, TAIPAN or LSST. Comment: 15 pages, 14 figures, 1 table. Accepted by MNRAS