Mohamed Afify
YOU?
Author Swipe
View article: Ecofriendly spectrophotometric techniques for the determination of imeglimin hydrochloride in pharmaceutical formulations
Ecofriendly spectrophotometric techniques for the determination of imeglimin hydrochloride in pharmaceutical formulations Open
View article: Simple and Sensitive Spectrophotometric Methods for Estimation of Vortioxetine HBr in Pharmaceutical Formulations
Simple and Sensitive Spectrophotometric Methods for Estimation of Vortioxetine HBr in Pharmaceutical Formulations Open
View article: Privacy-Preserving Transformers: SwiftKey's Differential Privacy Implementation
Privacy-Preserving Transformers: SwiftKey's Differential Privacy Implementation Open
In this paper we train a transformer using differential privacy (DP) for language modeling in SwiftKey. We run multiple experiments to balance the trade-off between the model size, run-time speed and accuracy. We show that we get small and…
View article: On-Device Emoji Classifier Trained with GPT-based Data Augmentation for a Mobile Keyboard
On-Device Emoji Classifier Trained with GPT-based Data Augmentation for a Mobile Keyboard Open
Emojis improve communication quality among smart-phone users that use mobile keyboards to exchange text. To predict emojis for users based on input text, we should consider the on-device low memory and time constraints, ensure that the on-…
View article: Simply Trainable Nearest Neighbour Machine Translation with GPU Inference
Simply Trainable Nearest Neighbour Machine Translation with GPU Inference Open
Nearest neighbor machine translation is a successful approach for fast domain adaption, which interpolates the pre-trained transformers with domain-specific token-level k-nearest-neighbor (kNN) retrieval without retraining. Despite kNN MT'…
View article: Early prognostication of ischemic stroke using computed tomography perfusion and S100 beta level
Early prognostication of ischemic stroke using computed tomography perfusion and S100 beta level Open
Background Early outcome prediction in ischemic stroke is a crucial concern for clinicians for proper decisions to reduce morbidity and mortality. The present study aimed to compare between computed tomography (CT) perfusion, National Inst…
View article: How Good Are GPT Models at Machine Translation? A Comprehensive Evaluation
How Good Are GPT Models at Machine Translation? A Comprehensive Evaluation Open
Generative Pre-trained Transformer (GPT) models have shown remarkable capabilities for natural language generation, but their performance for machine translation has not been thoroughly investigated. In this paper, we present a comprehensi…
View article: Domain Specific Sub-network for Multi-Domain Neural Machine Translation
Domain Specific Sub-network for Multi-Domain Neural Machine Translation Open
This paper presents Domain-Specific Sub-network (DoSS). It uses a set of masks obtained through pruning to define a sub-network for each domain and finetunes the sub-network parameters on domain data. This performs very closely and drastic…
View article: Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU
Fast Vocabulary Projection Method via Clustering for Multilingual Machine Translation on GPU Open
Multilingual Neural Machine Translation has been showing great success using transformer models. Deploying these models is challenging because they usually require large vocabulary (vocab) sizes for various languages. This limits the speed…
View article: Language Tokens: A Frustratingly Simple Approach Improves Zero-Shot Performance of Multilingual Translation
Language Tokens: A Frustratingly Simple Approach Improves Zero-Shot Performance of Multilingual Translation Open
This paper proposes a simple yet effective method to improve direct (X-to-Y) translation for both cases: zero-shot and when direct data is available. We modify the input tokens at both the encoder and decoder to include signals for the sou…
View article: Ensembling of Distilled Models from Multi-task Teachers for Constrained Resource Language Pairs
Ensembling of Distilled Models from Multi-task Teachers for Constrained Resource Language Pairs Open
This paper describes our submission to the constrained track of WMT21 shared news translation task. We focus on the three relatively low resource language pairs Bengali to and from Hindi, English to and from Hausa, and Xhosa to and from Zu…
View article: Ensembling of Distilled Models from Multi-task Teachers for Constrained\n Resource Language Pairs
Ensembling of Distilled Models from Multi-task Teachers for Constrained\n Resource Language Pairs Open
This paper describes our submission to the constrained track of WMT21 shared\nnews translation task. We focus on the three relatively low resource language\npairs Bengali to and from Hindi, English to and from Hausa, and Xhosa to and\nfrom…
View article: Score Combination for Improved Parallel Corpus Filtering for Low Resource Conditions
Score Combination for Improved Parallel Corpus Filtering for Low Resource Conditions Open
This paper describes our submission to the WMT20 sentence filtering task. We combine scores from (1) a custom LASER built for each source language, (2) a classifier built to distinguish positive and negative pairs by semantic alignment, an…
View article: Score Combination for Improved Parallel Corpus Filtering for Low\n Resource Conditions
Score Combination for Improved Parallel Corpus Filtering for Low\n Resource Conditions Open
This paper describes our submission to the WMT20 sentence filtering task. We\ncombine scores from (1) a custom LASER built for each source language, (2) a\nclassifier built to distinguish positive and negative pairs by semantic\nalignment,…
View article: Text-Independent Speaker Verification Based on Deep Neural Networks and Segmental Dynamic Time Warping
Text-Independent Speaker Verification Based on Deep Neural Networks and Segmental Dynamic Time Warping Open
In this paper we present a new method for text-independent speaker verification that combines segmental dynamic time warping (SDTW) and the d-vector approach. The d-vectors, generated from a feed forward deep neural network trained to dist…
View article: Text-Independent Speaker Verification Based on Deep Neural Networks and\n Segmental Dynamic Time Warping
Text-Independent Speaker Verification Based on Deep Neural Networks and\n Segmental Dynamic Time Warping Open
In this paper we present a new method for text-independent speaker\nverification that combines segmental dynamic time warping (SDTW) and the\nd-vector approach. The d-vectors, generated from a feed forward deep neural\nnetwork trained to d…
View article: Association of some adipokines and oxidative stress biomarkers with septic patients.
Association of some adipokines and oxidative stress biomarkers with septic patients. Open