Based on recent advances in natural language modeling and those in text generation capabilities, we propose a novel data augmentation method for text classification tasks. We use a powerful pre-trained neural network model to artificially synthesize new labeled data for supervised learning. We mainly focus on cases with scarce labeled data. Our method, referred to as language-model-based data augmentation (LAMBADA), involves fine-tuning a state-of-the-art language generator to a specific task through an initial training phase on the existing (usually small) labeled data. Using the fine-tuned model and given a class label, new sentences for the class are generated. Our process then filters these new sentences by using a classifier trained on the original data. In a series of experiments, we show that LAMBADA improves classifiers' performance on a variety of datasets. Moreover, LAMBADA significantly improves upon the state-of-the-art techniques for data augmentation, specifically those applicable to text classification tasks with little data.

In recent years, enterprise group chat collaboration tools, such as Slack, IBM»s Watson Workspace and Microsoft Teams, have presented unprecedented growth. With all the potential benefits of these tools - productivity increase and improved group communication - come significant challenges. Specifically, the 'always on' feature that makes it hard for users to cope with the load of conversational content and get up to speed after logging off for a while. In this demo, we present Collabot - a chat assistant service that implicitly learns users interests and social ties within a chat group and provides a personalized digest of missed content. Collabot assists users in coping with chat information overload by helping them understand the main topics discussed, collaborators, links and resources. This demo has two main contributions. First, we present a novel personalized group chat summarization algorithm; second the demonstration depicts a working implementation applied on different chat groups from different domains within IBM. A video, describing the demo can be found at https://www.youtube.com/watch?v=6cVsstiJ9vk.

MOTIVATION: Metabolic flux analysis (MFA) is a commonly used approach for quantifying metabolic fluxes based on tracking isotope labeling of metabolite within cells. Tandem mass-spectrometry (MS/MS) has been recently shown to be especially useful for MFA by providing rich information on metabolite positional labeling, measuring isotopic labeling patterns of collisional fragments. However, a major limitation in this approach is the requirement that the positional origin of atoms in a collisional fragment would be known a priori, which in many cases is difficult to determine. RESULTS: Here we show that MS/MS data could also be used to improve flux inference even when the positional origin of fragments is unknown. We develop a novel method, metabolic flux analysis/unknown fragments, that extends on standard MFA and jointly searches for the most likely metabolic fluxes together with the most plausible position of collisional fragments that would optimally match measured MS/MS data. MFA/UF is shown to markedly improve flux prediction accuracy in a simulation model of gluconeogenesis and using experimental MS/MS data in Bacillus subtilis.

MOTIVATION: Computational modeling in metabolic engineering involves the prediction of genetic manipulations that would lead to optimized microbial strains, maximizing the production rate of chemicals of interest. Various computational methods are based on constraint-based modeling, which enables to anticipate the effect of genetic manipulations on cellular metabolism considering a genome-scale metabolic network. However, current methods do not account for the presence of competing pathways in a metabolic network that may diverge metabolic flux away from producing a required chemical, resulting in lower (or even zero) chemical production rates in reality-making these methods somewhat over optimistic. RESULTS: In this article, we describe a novel constraint-based method called RobustKnock that predicts gene deletion strategies that lead to the over-production of chemicals of interest, by accounting for the presence of competing pathways in the network. We describe results of applying RobustKnock to Escherichia coli's metabolic network towards the production of various chemicals, demonstrating its ability to provide more robust predictions than those obtained via current state-of-the-art methods.