Streaming Anchor Loss: Augmenting Supervision with Temporal Significance

Streaming neural network models for fast frame-wise responses to various speech and sensory signals are widely adopted on resource-constrained platforms. Hence, increasing the learning capacity of such streaming models (i.e., by adding more parameters) to improve the predictive power may not be viable for real-world tasks. In this work, we propose a new loss, Streaming Anchor Loss (SAL), to better utilize the given learning capacity by encouraging the model to learn more from essential frames. More specifically, our SAL and its focal variations dynamically modulate the frame-wise cross entropy loss based on the importance of the corresponding frames so that a higher loss penalty is assigned for frames within the temporal proximity of semantically critical events. Therefore, our loss ensures that the model training focuses on predicting the relatively rare but task-relevant frames. Experimental results with standard lightweight convolutional and recurrent streaming networks on three different speech based detection tasks demonstrate that SAL enables the model to learn the overall task more effectively with improved accuracy and latency, without any additional data, model parameters, or architectural changes.

Blockchain-based decentralized voting system security Perspective: Safe and secure for digital voting system

This research study focuses primarily on Block-Chain-based voting systems, which facilitate participation in and administration of voting for voters, candidates, and officials. Because we used Block-Chain in the backend, which enables everyone to trace vote fraud, our system is incredibly safe. This paper approach any unique identification the Aadhar Card number or an OTP will be generated then user can utilise the voting system to cast his/her vote. A proposal for Bit-coin, a virtual currency system that is decided by a central authority for producing money, transferring ownership, and validating transactions, included the peer-to-peer network in a Block-Chain system, the ledger is duplicated across several, identical databases which is hosted and updated by a different process and all other nodes are updated concurrently if changes made to one node and a transaction occurs, the records of the values and assets are permanently exchanged, Only the user and the system need to be verified no other authentication required. If any transaction carried out on a block chain-based system would be settled in a matter of seconds while still being safe, verifiable, and transparent. Although block-chain technology is the foundation for Bitcoin and other digital currencies but also it may be applied widely to greatly reduce difficulties in many other sectors, Voting is the sector that is battling from a lack of security, centralized-authority, management-issues, and many more despite the fact that transactions are kept in a distributed and safe fashion.

Enhancing Slot Tagging with Intent Features for Task Oriented Natural Language Understanding using BERT

Recent joint intent detection and slot tagging models have seen improved performance when compared to individual models. In many real-world datasets, the slot labels and values have a strong correlation with their intent labels. In such cases, the intent label information may act as a useful feature to the slot tagging model. In this paper, we examine the effect of leveraging intent label features through 3 techniques in the slot tagging task of joint intent and slot detection models. We evaluate our techniques on benchmark spoken language datasets SNIPS and ATIS, as well as over a large private Bixby dataset and observe an improved slot-tagging performance over state-of-the-art models.