Kevin Patel

Are Word Embedding-based Features Useful for Sarcasm Detection?

This paper makes a simple increment to state-of-the-art in sarcasm detection research. Existing approaches are unable to capture subtle forms of context incongruity which lies at the heart of sarcasm. We explore if prior work can be enhanced using semantic similarity/discordance between word embeddings. We augment word embedding-based features to four feature sets reported in the past. We also experiment with four types of word embeddings. We observe an improvement in sarcasm detection, irrespective of the word embedding used or the original feature set to which our features are augmented. For example, this augmentation results in an improvement in F-score of around 4\% for three out of these four feature sets, and a minor degradation in case of the fourth, when Word2Vec embeddings are used. Finally, a comparison of the four embeddings shows that Word2Vec and dependency weight-based features outperform LSA and GloVe, in terms of their benefit to sarcasm detection.

Evaluation of applicability of modified vector space representation for in-VM malicious activity detection in Cloud

Malware writers use increasingly complex evasion mechanisms to ensure the concealment of malware against standard anti-malware suites. To identify malware through its behaviour, rather than its approach is an interesting venue of exploration. System call traces are highly indicative of a process behaviour. However, it is difficult to acquire system calls of all processes running on a physical machine. Fortunately, the same cannot be said for the virtual machines, owing to the advancement of Virtual Machine Introspection (VMI) techniques. This opens up the possibility of utilizing system call information for malicious activity detection. In this paper, we study different representations of system call information and evaluate their applicability for in- VM malicious activity detection in Cloud environment.

ExplicitPRISMSymm: Symmetry Reduction Technique for Explicit Models in PRISM

Probabilistic model checking of concurrent system involves exhaustive search of the reachable state space associated with the system model. Symmetry reduction is a commonly employed technique that enables model checking of exponentially large models. Most work on symmetry reduction focuses on symbolically represented probabilistic models, which are easy to build and perform reasonably well at property checking. In this work, we rather focus on explicitly represented probabilistic models. We report that explicitly represented models perform well at property checking, but face hurdles in model construction. We present an on-the-fly symmetry reduction technique for explicitly represented models. It significantly reduces build time and thus explicit model representation as an efficient alternative to symbolic model representation.

Adapting Pre-trained Word Embeddings For Use In Medical Coding

Word embeddings are a crucial component in modern NLP. Pre-trained embeddings released by different groups have been a major reason for their popularity. However, they are trained on generic corpora, which limits their direct use for domain specific tasks. In this paper, we propose a method to add task specific information to pre-trained word embeddings. Such information can improve their utility. We add information from medical coding data, as well as the first level from the hierarchy of ICD-10 medical code set to different pre-trained word embeddings. We adapt CBOW algorithm from the word2vec package for our purpose. We evaluated our approach on five different pre-trained word embeddings. Both the original word embeddings, and their modified versions (the ones with added information) were used for automated review of medical coding. The modified word embeddings give an improvement in f-score by 1% on the 5-fold evaluation on a private medical claims dataset. Our results show that adding extra information is possible and beneficial for the task at hand.

On-the-Fly Symmetry Reduction of Explicitly Represented Probabilistic Models

Quantitative analysis of concurrent systems becomes intract-able due to searches over the enormous state space. Since these systems often contain many identical processes consisting of symmetrical and interchangeable components, this problem can be tackled using symmetry reduction. In this paper, we present an on-the-fly symmetry reduction technique that is applicable to explicitly represented models in the probabilistic setting. We have performed the experiments by integrating our technique into PRISM probabilistic model checker. Experimental results are very encouraging with considerable reductions in both the time taken for property evaluation and the associated memory usage.