Search query understanding (QU) is an important building block of the modern e-commerce search engines. QU extracts multiple intents from customer queries, including intended color, brand, etc. One of the most important tasks in QU is predicting which product category the user is interested in. In our work we are tapping into query product type classification (Q2PT) task. Compared to classification of full-fledged texts, Q2PT is more complicated because of the ambiguity of short search queries, which is aggravated by language and cultural differences in worldwide online stores. Moreover, the span and variety of product categories in modern marketplaces pose a significant challenge. We focus on Q2PT inference in the global multi-locale e-commerce markets, which need to deliver high quality user experience in both large and small local stores alike. The common approach of training Q2PT models for each locale separately shows significant performance drops in low-resource stores and prevents from easily expanding to a new country, where the Q2PT model has to be created from scratch. We use transfer learning to address this challenge, augmenting low-resource locales through the vast knowledge of the high-resource ones. We introduce a unified, locale-aware Q2PT model, sharing training data and model structure across worldwide stores. We show that the proposed unified locale-aware Q2PT model has superior performance over the alternatives by conducting extensive quantitative and qualitative analysis on the large-scale multilingual e-commerce dataset across 20 worldwide locales. Our online A/B tests have shown that using locale-aware model improves over the previous user experience, increasing customer satisfaction.

Performing large-scale malware classification is increasingly becoming a critical step in malware analytics as the number and variety of malware samples is rapidly growing. Statistical machine learning constitutes an appealing method to cope with this increase as it can use mathematical tools to extract information out of large-scale datasets and produce interpretable models. This has motivated a surge of scientific work in developing machine learning methods for detection and classification of malicious executables. However, an optimal method for extracting the most informative features for different malware families, with the final goal of malware classification, is yet to be found. Fortunately, neural networks have evolved to the state that they can surpass the limitations of other methods in terms of hierarchical feature extraction. Consequently, neural networks can now offer superior classification accuracy in many domains such as computer vision and natural language processing. In this paper, we transfer the performance improvements achieved in the area of neural networks to model the execution sequences of disassembled malicious binaries. We implement a neural network that consists of convolutional and feedforward neural constructs. This architecture embodies a hierarchical feature extraction approach that combines convolution of n-grams of instructions with plain vectorization of features derived from the headers of the Portable Executable (PE) files. Our evaluation results demonstrate that our approach outperforms baseline methods, such as simple Feedforward Neural Networks and Support Vector Machines, as we achieve 93% on precision and recall, even in case of obfuscations in the data.