Quantum Computing and Security Aspects of Attention‐Based Visual Question Answering with Long Short‐Term Memory
Madhav Shrivastava, Rajat Patil, Vivek Bhardwaj, Romil Rawat, Shrikant Telang, Anjali Rawat · 2023
In this write-up, we will study the core concept of VQA the LSTM with Att.-based models and CNN Att. models that combine the local images’ hidden features and the answer of the question which is raised by end user is produced from the portion of the image which is generated by image dataset. So, the word Att. means that it only keeps Att. on those parts which are relevant to both object and keywords in the question. We are not considering the outlier to reduce the chances of mistakes. To combine the results from the image and given questions we are using multi-layer awareness. In this proposal of QC in field of VQA and LSTM, we tried to use this concept of MM Networks and presented our view on vulnerability of a primary/novel kind of attack that we call as DKMB. This hard kind of theft breaks the complex fusion (Combinations) mechanism take into consideration by a prime state-of-art networks to fuse BDs which are both effective, efficient, and stealthy. Here, we are proposing a multi-model for VQA with Att.-Based LSTM along with loopholes where attacker can attack and influence system which can be tackled with Quantum Computing and Cybersecurity Concepts.