Abstract

In AI, the ability of intelligent agent to model human player in games such as Backgammon, Chess and Go has been an important metric in benchmarking progress. Fundamentally, the games mentioned above can be characterized as competitive and zero-sum. In contrast, games such as Pictionary and Dumb Charades falls into the category of ‘social’ games. Unlike competitive games, the emphasis is on cooperative and co-adaptive game-play in a relaxed setting. Such social games can form the basis for the next wave of game-driven progress in AI. Pictionary™ is a wonderful example of cooperative game play to achieve a shared goal in communication-restricted settings. This popular sketch-based guessing game, which we employ as a use case, provides an opportunity to analyze shared goal cooperative game play in restricted communication settings. To enable the study of Pictionary and to understand various aspects associated with the game play, we designed a software ecosystem for web-based online game of Pictionary dubbed PICTGUESS. To overcome several technological and logistic barriers, which the actual game presents, we implemented a simplified setting for PICTGUESS wherein a game consists of a time-limited episode involving two players - a Drawer and a Guesser. The Drawer is tasked with conveying a given target phrase to a counterpart Guesser by sketching on a whiteboard within that time limit. However, occasionally some players in Pictionary draw atypical sketch content. While such content is occasionally relevant in the game context, it sometimes represents a rule violation and impairs the game experience. To address such situations in a timely and scalable manner, we introduce DRAWMON, a novel distributed framework for automatic detection of atypical sketch content in concurrently occurring Pictionary game sessions. We build specialized online interfaces to annotate atypical sketch content, resulting in ATYPICT, the first ever atypical sketch content dataset. We use ATYPICT to train CANVASNET, a deep neural atypical content detection network. We utilize CANVASNET as a core component of DRAWMON. Our analysis of post deployment game session data indicates DRAWMON’s effectiveness for scalable monitoring and atypical sketch content detection. Beyond Pictionary, our contributions can also serve as a design guide for customized atypical content response systems involving shared and interactive whiteboards.

Abstract

River water quality is a vital component of the ecosystem which is the main source for drinking, irrigation and more. In recent years, increasing urban growth and expansion have resulted in structural deterioration and functional degradation of river systems, resulting in significant volumes of harmful contaminants being discharged into the water. River pollution substantially impedes long-term economic and social growth and endangers human health. In order to enhance the water environment, it is essential to examine the reasons for water quality deterioration. River water quality modeling is the finest way to assess and monitor water quality. River water quality models are mathematical models that simulate the behavior of pollutants in surface water bodies used to predict the concentration of pollutants in the water and assess the overall river water quality. QUAL2K is a widely-used water quality model that can simulate water quality parameters, such as Dissolved Oxygen (DO), nutrients, and pH, in streams and rivers. However, simulation of many water quality parameters through water quality model does not assess the status and overall pollution extent of any river system. Single river water quality index is required for evaluation of overall pollution extent and to adopt necessary water quality control measures. Therefore, a holistic approach which can integrate river water quality simulation model and pollution extent characterization model is necessary. Thus, this study aims to develop an integrated, holistic model approach to estimate the pollution extent in terms of river water quality index utilizing various river water quality model parameters. Furthermore, the study aimed to analyze the spatial variability of river pollution extent using Geographic Information System. In this present study, the river water quality model, QUAL2K, is used to assess the quality of Bhadra river stretch, one of the major tributaries of the Tunga-Bhadra River situated in Karnataka. The study stretch considered is around 27 km which is divided into three reaches with elements of 1km as 3,4,20 for each reach respectively. In this current study, we analysed the effects of wastewater discharge from the monitoring stations such as the Mysore Paper Mill (MPM), the Visveshvarya Industrial Steel Limited (VISL), and the Bhadravathi city to simulate the Dissolved Oxygen by varying the Biochemical Oxygen Demand (BOD) loads (25%, 50%,70%,100%) coming from different pollutant sources within the study stretch. The observed water quality parameter data from 2006 to 2017 for those monitoring stations has been obtained from Advanced Centre for Integrated Water Resource (ACIWR). The water quality parameters like flow rate and pollution point sources discharge are the highly sensitive water quality parameters for modelling the QUAL2K model. There must be a reduction of 25% of BOD effluent to reach the minimum standards set by the Central Pollution Control Board (CPCB). It is noted that a 75% reduction of BOD effluent from point sources will lead to an increase of 15% average DO throughout the study stretch. The weighted Average Water Quality Index (WAWQI) method is used in estimating Water Quality Index (WQI) by using QUAL2K simulation results for the study stretch. QUAL2K model was calibrated from April 2006 to October 2013 and validated from November 2013 to March 2017. The estimated WQI values range from 92.35 to 112, indicating the quality classification ranges from very poor to unfit for consumption for the Bhadra river. It was observed that the quality status of Bhadra river water was very poor in upstream and downstream segments, while unfit for consumption in the middle segment indicating access to industrial and anthropogenic activities. For spatial analysis, Arc-Geographical Informal System (GIS) software and Inverse Distance Weighted (IDW) interpolation technique is used to generate water quality parameters concentration level and WQI maps. In context to this, it was found that the middle segment of Bhadra study stretch falls into the category of unfit for consumption due to the combination of effluents from industries like MPM and VISL and domestic effluents from Bhadravathi city. Overall, this study revealed that the MPM water quality station was a hotspot for river water quality degradation. Based on all these analyses, we provide a framework for the integration of the three components, i.e., QUAL2K calibration, WQI and spatial visualization for water quality management and policymakers to take decisions about water quality management in Bhadra river stretch .

Abstract

In this work, we develop three (related) schemes to generate novel molecules based on a seed molecule using an attentive captioning network. Input is the grid representation of the seed molecule, which could be one of the following: voxelised grid of the seed molecule, a grid reconstructed grid from it, or a sampled grid from it. The reconstructed grid is generated by passing the voxelised grid to the variational autoencoder and the sampled grid is generated by conditioning the decoder phase of the variational autoencoder on pharmacophoric requirements. The first scheme called ‘Direct Generation’ uses a RNN with the voxelised grid of the seed molecule as input to give an SMILES output of generated molecule. The second and third schemes utilize an additional variational autoencoder (VAE) to generate the grid which is used as input to RNN mentioned above in the subsequent step; the latent space of this VAE is modelled as a Riemannian manifold attached with a metric which is learnt along with the encoder and decoder networks of this VAE, which we name RHVAE. The second scheme, named ‘Autoencoded Generation’, takes as input the seed molecules’ voxelised grid representation for the encoder and its output along with the Remianian metric generates the latent space representation; this latent space representation along with the pharmacophoric requirement conditions form inputs for the decoder which outputs the ‘reconstruction grid’. In the third scheme, named ‘Sampled Generation’, starts with a point (and its Remanian metric) randomly sampled from latent space distribution learnt during training. This is evolved using either Hamiltonain Monte Carlo or Random Walk Monte Carlo, and then the evolved point with pharmacophoric conditions is sent to decoder to generate a ‘sampled grid’. In the subsequent step of Autoencoded Generation (Sampled Generation), this reconstructed grid (sampled grid) is sent to RNN for obtaining the SMILES of generated molecules. Overall, we demonstrate the generation of meaningful ligand shapes through the autoencoder network which can then be passed to our attentive captioning network to generate novel molecules while requiring smaller data sets for training while retaining the similar performance.

Abstract

Automated text generation for low resource (LR) languages is a critical area of research because of lack of contributors to encyclopedic texts, notably on Wikipedia. The majority of the work done so far on Wikipedia text generation has concentrated on creating English-only Wikipedia articles by summarizing English reference articles. Monolingual text generation is unable to address this issue for low-resource languages due to the lack of reference materials. To start addressing these problems, we propose a benchmark dataset called XWikiRef that consists of ∼69K in Wikipedia articles from five different domains and eight different languages. Utilizing this dataset, we train a two-stage system that outputs a section-specific LR summary from an input of a set of citations and a section title. One crucial aspect of content organization is the creation of article outlines, which summarize the primary topics and subtopics covered in an article in a structured manner. We introduce a pipeline called XOutlineGen, which generates cross-lingual outlines for encyclopedic texts from reference articles. XOutlineGen uses the XWikiRef dataset, which consists of encyclopedic texts generated from reference articles and section titles. Our pipeline employs this dataset to train a two-step generation model, which takes the article title and set of references as inputs and produces the article outline. Commonsense question-answering (QA) methods combine the power of pre-trained Language Models (LM) with the reasoning provided by Knowledge Graphs (KG). A typical approach collects nodes relevant to the QA pair from a KG to form a Working Graph (WG) followed by reasoning using Graph Neural Networks (GNNs). This faces two major challenges: (i) it is difficult to capture all the information from the QA in the WG, and (ii) the WG contains some irrelevant nodes from the KG. To address these, we propose GrapeQA with two simple improvements on the WG: (i) Prominent Entities for Graph Augmentation identifies relevant text chunks from the QA pair and augments the WG with corresponding latent representations from the LM, and (ii) Context-Aware Node Pruning removes nodes that are less relevant to the QA pair. We evaluate our results on OpenBookQA, CommonsenseQA and MedQA-USMLE and see that GrapeQA shows consistent improvements over its LM + KG predecessor (QAGNN in particular) and large improvements on OpenBookQA. We utilize the idea of relevance scoring from this work in our next work which is called XWikiGen for performing neural extractive summarization With this study, we propose XWikiGen, a task of cross-lingual multi-document summarization of text from numerous reference articles written in different languages to produce Wikipedia-style material. The suggested approach is built on the novel idea of using neural unsupervised extractive summarization to roughly select salient information and then using a neural abstractive model to produce the section-specific text. Extensive experiments have revealed that multi-domain training generally outperforms a multi-lingual and multi-lingualmulti-domain perform best, even better then previous two settings. Overall, we propose a new dataset called XWikiRef for the task of encyclopedic text generation, a 2 stage pipeline XOutlineGen to generate article outline from references and a cross-lingual multi-document summarization based 2 stage pipeline XWikiGen to generate Wikipedia style text. Along with these, we also explore the idea of relevance scoring first in the domain of question answering with reasoning (GrapeQA) and then in the context of unsupervised extractive summarization.

Abstract

Metaverse has gained massive popularity these days owing to its potential in various industries and research fields such as the health sector, online social interaction, gaming and entertainment, education, eCommerce transactions, etc. Some of the challenges to be addressed in building a Metaverse, including human-computer interactions and scene understanding, rely on AI techniques for solutions. However, most of these methods rely on supervised learning that requires labor-intense annotated data and have limited generalizability. Self-supervised learning methods offer a viable alternative to overcome these limitations as well as leverage large amounts of readily available unlabeled raw data, which is often more abundant and easier to collect. These methods can learn to capture temporal or spatial relationships within data, modeling contextual information, which is valuable for several computer vision tasks. Hence, these methods can be useful in addressing the challenges of building the Metaverse. Within the Metaverse, users can interact with each other and a digital environment through avatars. Using UV parameterized maps of avatars, textures can be accurately applied, resulting in realistic skin, clothing, and other visual details. They also play a crucial role in the creation of virtual fashion and design within the Metaverse. This can also benefit the E-Commerce industry, particularly fashion and apparel, which have experienced tremendous growth in recent years. The Metaverse can provide immersive virtual environments for virtual try-on where users can explore and interact with products more realistically and engagingly than traditional online shopping. Given the above advantages, we intend to address the following two challenges in this thesis. First, we explore self-supervised data-driven methods for UV parameterization of general objects. The existing methods for surface parameterization of arbitrary 3D objects face challenges when dealing with closed surfaces and regions of extreme extrinsic curvature. Mapping a surface from 3D to 2D almost always introduces a certain amount of distortion, and the aim is to keep this distortion as low as possible. Finding optimal seams that lead to low distortion is another challenge. We present a novel framework for learning the discretization-agnostic surface parameterization of arbitrary 3D objects with closed and open surfaces. We evaluate our framework on multiple 3D objects from the publicly available dataset and report a comparison with conventional methods. Secondly, utilizing self-supervised methods, we aim to innovate a solution for a 3D virtual tryon system. The goal is to retarget real, non-parametric garment meshes over a target human body (parametric or non-parametric). This 3D virtual try-on system needs to generalize to arbitrary body shapes and poses, modeling topological differences among various categories of garments, along with realistic deformations arising out of the physical interaction with the underlying body and resolving the penetration/intersection of the garment with the underlying body. These systems need to run in real time with very less delay. We propose a self-supervised method for draping non-parametric, 3D garment meshes by first obtaining the initial alignment between the garment and the human body by establishing correspondences via Isomap Embeddings. Further, this coarse retargeting is refined by training an MLP that preserves the geometry of the garments guided by our novel losses. We propose a wrinkle generation module to obtain realistic details on the draped garments. We also contribute a new dataset of real-world reposed garments with realistic noise and topological deformations. Finally, we discuss the limitations of our work and lay down the potential solutions that can be explored. We also discuss the future directions that can be pursued based on the findings of our work. We believe this thesis advances the field of virtual try-on systems significantly while providing a learningbased solution for parameterization.

Abstract

Genome integrity is seriously threatened by UV radiation’s ability to cause damage in DNA. CPD and 6-4PP are two of the most frequent lesions brought on by Ultraviolet radiation. CPD is a DNA lesion occurring from the covalent linking of nearby pyrimidine bases through their C5 and C6 carbons, generating a cyclobutane ring. In contrast, 6-4PP develops when neighbouring pyrimidine bases in DNA become crosslinked following UV exposure, producing a covalent bond between their C6 and C4 carbons. If left untreated, these lesions can result in mutations and even the development of cancer. Clarifying the DNA repair process depends on our ability to comprehend the mechanisms underpinning DNA lesion identification. DNA damage repair proteins such as XPC/Rad4 recognize and facilitate repair of these damages. In order to better understand how XPC/Rad4 recognises DNA lesions in Nucleotide Excision Repair (NER), this thesis will focus on how it interacts with CPD and 6-4PP. The highly conserved repair process known as NER is in charge of locating and eradicating diverse DNA damages. The current thesis looks at the mechanism, energetics and dynamics of Rad4 sequence-specific lesion recognition. The flipping of a pair of partner bases that are present opposite to the lesion (CPD or 64PP) is one of the crucial biochemical steps that take place while XPC or Rad4 tries to identify and bind to a DNA lesion. Major events in NER like insertion of BHD3-β hairpin, flipping of partner bases and Rad4/XPC association with the damaged DNA were studied by formulating relevant reaction coordinates or Collective Variables. These CVs were then used in Molecular Dynamics and enhanced sampling method, Umbrella Sampling to produce energy profiles for the aforementioned events. A deeper understanding of these energy profiles reveal that insertion of BHD3-β hairpin takes place only after the flipping of partner bases has taken place and a cavity is induced in the active site of the DNA but 6-4PP containing complex promotes the insertion of BHD3-β hairpin with higher ease and flexibility compared to a CPD containing complex.This study also reveals that flipping of 5’-dA partner base takes place before flipping of 3’-dA partner base. And these observations were consistent for both lesions, CPD and 64PP. The results of this study will be useful to improve our knowledge of the molecular processes behind Rad4’s identification of DNA lesions and give light on the critical steps in the NER pathway. Finally, this information may aid in the creation of focused therapeutic approaches for preventing and treating UV-induced DNA damage, enhancing our capacity to protect genomic integrity and lessen the dangers of UV radiation exposure

Abstract

The semiconductor industry is driven by the need to design smaller, faster and low power con- suming circuits. A comparator is an integral part of any electronic system and by default the comparators exhibit hysteresis phenomenon. A Schmitt trigger is the most commonly used comparator in circuit design. Rigorous efforts have been made to speed up the circuit perfor- mance by various techniques at the system, circuit and transistor level and they have demon- strated incremental improvement. However, all these efforts are focused on reducing the switch- ing time between the logic levels and shortening the width of hysteretic loop while maintaining noise immunity. This thesis presents a new paradigm shift from the conventional techniques of circuit speed up wherein the hysteretic device itself is replaced with a proteretic one and efforts to improve the circuit speed have been demonstrated successfully. Hysteresis has an inherent switching delay, whereas, proteresis or inverse hysteresis demonstrates an early response to the switching action of the circuit. Proteresis is a known phenomenon across various domains like pharmacokinetics (PK) and pharmacodynamics (PD) drugs, ferro-electric materials and optical bi-stable devices, but their impact and study in area of Integrated circuits has been very limited. This is because proteresis is induced by feed forwarding the input to couple of hysteretic devices (Schmitt trigger) and this leads to output of circuit being stable for a small input ranges and prevalent instability over the remaining input range. However, this thesis proposes a novel mechanism of stabilizing the proteretic output over a wide input voltage range and hence leverages the benefit of early switching characteristics for improving the circuit speed by a factor of 25%. A detailed mathematical model for proteretic circuit design over various input ranges is presented and the trade-off in terms of area and power is also discussed with the ramp generator as a design example. Hysteretic and proteretic devices are bi-stable and their interstate switching depends on triggering voltages. Conventionally, devices can be classified to work as either hysteretic or proteretic. This thesis proposes the first attempt to describe a circuit that switches between two opposite characteristics of hysteresis and proteresis to present a new circuit called Prohys switch. The output of prohys switch demonstrates a limited amount of randomness in its first cycle of operation and this factor is a strong reason to employ a prohys switch for designing a Physical unclonable Function (PUF). In recent times PUF’s has gained immense popularity for securing the IC’s by providing unique identification code to each chip. The key design parameters of a PUF are uniqueness and reliability and designing a highly efficient PUF with optimal values of uniqueness and reliability is quite a challenge. Reliability depends on the chip’s ability to resist changes to supply voltage and temperature variations, whereas, uniqueness depends on process variations during chip fabrication. Multiple PUF designs that employ reliability enhancement circuits and security algorithms achieve these design characteristics. Nonetheless, these techniques are design overheads. Finally, this thesis presents a novel PUF based on the proHys switch called the prohys PUF. The proposed prohys PUF befittingly satisfies both uniqueness and reliability criteria, without any additional circuitry or security algorithms.

Abstract

The safety improvements in the medical systems or devices, specifically non-invasive patient monitoring systems (PMS) and point-of-care (POC) devices of human health monitoring (HHMS) systems, are in great need to perform precise measurements of vital parameters with uninterrupted continuous monitoring for diagnosis of significant human health ailments. During critical or non-critical nursing times, frequent monitoring of short and long-duration measures of vital parameters (like ECG, EEG, Respiratory, SpO2, Blood-pressure, Temperature, etc.) and non-vital parameters (like Glucose levels, Hemoglobin, Urea in blood, etc.) will allow for better nursing and early detection of diseases like cardiovascular diseases, respiratory diseases, liver dysfunction, diabetes, renal diseases, and other psychological disorders. A non-invasive medical instrumentation system for accomplishing these tasks must meet several criteria like accuracy and precision of the measurement, reduction of false alarms, effective detection of faults, and fault tolerance to systemic or random failures with an uninterrupted performance during nursing times. Many research groups have made many efforts worldwide to make strategies and guidelines and standardize the procedures for medical systems based on safety design approaches to raise alerts for any deviance during non-invasive monitoring of health parameters. In addition, health monitoring instruments should be portable, low-cost, and reliable over time. Many modes of diagnosis and related instrument maintenance procedures satisfy these criteria. However, most of them are still in the improvement stage for developing resilient instruments for integrating with critical auto-robotic surgery instruments with minimal human intervention. Several approaches have been proposed, out of which safety-related design approaches like using 2oo2, 2oo3, 1oo2, etc., along with AI-based data analytics, have the advantage of meeting these rigorous demands in fundamental safety improvements of Medical Systems. Based on the safety-related design 2oo2 concept, a configurable system prototype of the cardiac health monitoring system (CHMS) is developed and evaluated to meet the set objectives, such as fault detection effectiveness and fault tolerance with improved safety configurability. This configurable system uses various sensors to collect the bio-medical data in parallel. Primarily, three diverse sensors are used non-invasively in sensing the bio-signals in different forms like electric potential, light, and sound signals for computations. These diverse sensors are used to detect biomedical signals to obtain data from electrocardiogram (ECG), Photo-plethysmogram (PPG), and Phonocardiogram (PCG). Therefore, the accuracy of the vital estimates and the fundamental safety improvements were evaluated using this multimodal system with AI-based fault detection and predictive maintenance techniques. Traditional statistical and AI-based techniques have acquired authentic measurements of human health parameters from diverse signals received simultaneously from various sensing circuits like PPG/ECG/EEG. Secondly, these bio-signals are calibrated independently with known algorithms in diverse. Finally, these obtained parameters, along with the built-in-test (BIT) system for health signals, are processed with implemented safety functions, and the algorithms to generate the correct human health parameters and prognosticate abnormalities of human health 200 patients were available for instrument evaluation trials for HR parameter monitoring and tested after taking informed consent. A MATLAB-based CHMS tool is developed for configurable and then implemented on a field-programmable gate array (FPGA) to minimize the designed circuitry with improved resiliency of the instrument system. The CHMS has been configured to 2oo2 with the selected HR parameter to estimate the system's availability and health. The collected HR output was subjected to data analytics against individually collected data. We found a significant reduction in the generation of unimportant alarms and increased uninterruptable System availability by 45% to 55%, along with normal and abnormal artifact data. The measured normal artifacts are more than 99% accurate and are used for prognosis. The abnormal data is used for edge-AI-based analytics to infer the system's health for prescriptive maintenance. An experimental study has been carried out to effectively segregate normal and abnormal signals in 2oo2 and 2oo3 configurations. A detailed analysis is carried out in various sensor configurations as proposed. Similarly, in the 2oo3 configuration, we significantly improved the system's availability from 55% to 95% by eliminating spurious alarms with reduced downtime and improved accurate data vital parameters. Further, a reliability assessment is performed on CHMS on identified parameters, such as measuring Avai

Abstract

Blockchain and Reinforcement Learning (RL) are two game-changing research areas that have received a lot of attention recently. In recent years, significant advances in RL have resulted in tremendous success in solving various sequential decision-making problems in machine learning. The two most successful RL applications are discussed in this work, unmanned aerial vehicles, and blockchain. Blockchain is a distributed ledger technology with its first application in Bitcoin. The main challenge in blockchain-based cryptocurrencies is to provide a distributed trust environment with high security like in a centralized financial system. The current throughput of Bitcoin is around 4 to 7 transactions per second and confirmation latency is about one hour. If Bitcoin has to go mainstream, the throughput has to be in the order of thousands of transactions per second with very low latency in the order of a few seconds. Recent advances in blockchain research proposed consensus algorithms that scale bitcoin, such as sharding and Prism-based blockchain. However, the security of Bitcoin is very high that it can tolerate up to 50% adversarial nodes and avoids double spending attacks. The current blockchain size is over 260 GB and is growing at an astonishing rate imposing a huge storage requirement on the nodes. Recent developments improving the Bitcoin consensus have shown that there is a tradeoff between decentralization, scaling, and security. In order to scale blockchain, we leverage coding theory and RL in this thesis. Due to the increasing storage requirement for blockchains, the computation can be afforded by only a few miners. Sharding has been proposed to scale blockchains so that the storage and transaction efficiency of the blockchain improves at the cost of a security guarantee. Incorporating coding theory into existing consensus algorithms has demonstrated improvements in terms of storage efficiency and low latency. A Secure-Repair-Blockchain (SRB) is proposed which aims to decrease the storage cost at the miners. In addition, SRB also decreases the bootstrapping cost, which allows for new miners to easily join a sharded blockchain. In order to reduce storage, coding-theoretic techniques are used in SRB. In order to decrease the amount of data that is transferred to the new node joining a shard, the concept of exact repair secure regenerating codes is used. The proposed blockchain protocol achieves lower storage than those that do not use coding and achieves lower bootstrapping costs as compared to the different baselines. Prism is a recent blockchain algorithm that achieves the physical limit on throughput and latency without compromising security. However, like the traditional blockchain systems, Prism also has a trade-off between security, latency, and cost. In recent days, reinforcement learning approaches are investigated in traditional blockchains, to improve performance. In this work, we apply Deep Reinforcement Learning (DRL) to one of the promising blockchain protocols, Prism, to optimize its performance. We propose a Deep Reinforcement Learning-based Prism Blockchain (DRLPB) scheme which dynamically optimizes the parameters of Prism blockchain and helps in achieving a better performance. In DRLPB, we apply two widely used DRL algorithms, Dueling Deep Q Networks (DDQN) and Proximal Policy Optimization (PPO). This work presents a novel approach to applying DDQN and PPO to a blockchain protocol and comparing the performance. The analysis of Prism in terms of latency, and security level considering other blockchain parameters is provided. Using the analysis, the DRLPB scheme adapts the Prism blockchain parameters to enhance the security upto 84% more than Prism, while still preserving the performance guarantees of Prism. The recent advancements in the field of Internet of Things (IoT) motivate the development of a secure infrastructure for storing and sharing vast amounts of data. Blockchain, a distributed and immutable ledger, is best known as a potential solution to data security and privacy for IoT. The scalability of blockchain, which should optimize the throughput and handle the dynamics of the IoT environment, becomes a challenge due to the enormous amount of IoT data. The critical challenge in scaling blockchain is to guarantee decentralization, latency, and security of the system while optimizing the transaction throughput. this paper presents a deep reinforcement learning (DRL)-based performance optimization for blockchain-enabled IoT. We consider one of the recent promising blockchains, Prism as the underlying blockchain system because of its good performance guarantees. We integrate the IoT data to Prism Blockchain and optimize the performance of the system by leveraging Proximal Policy Optimization (PPO) method. The DRL method helps to optimize the blockchain parameters like mining rate and mined blocks to adapt to the environment dynamics of

Abstract

Hardware accelerators are being designed to offload compute-intensive tasks such as deep neural networks from the CPU to improve the overall performance of an application, specifically on the performance-per-watt metric. With the evolution of speech recognition based applications, many deep learning models for Automatic Speech Recognition have been proposed. Encoder-decoder-based sequenceto-sequence models such as the Transformer model have demonstrated state-of-the-art results in end-toend automatic speech recognition systems (ASRs). However, the Transformer model being intensive on memory and computation poses a challenge for an FPGA implementation. This thesis proposes an end-to-end architecture to accelerate a Transformer for an ASR system. The host CPU orchestrates the computations from different encoder and decoder stages of the Transformer architecture on the designed hardware accelerator with no necessity for intervening FPGA reconfiguration. The intermediate stages, like data pre-processing and feature extraction, are performed on the host while the complex recognizer, i.e., the Transformer model, is offloaded onto an FPGA. The communication latency is hidden by prefetching the weights of the next encoder/decoder block while the current block is being processed. The larger computations in the model are split across both the Super Logic Regions (SLRs) of the FPGA, mitigating the inter-SLR communication. The proposed design presents an optimal latency, exploiting the available resources. The accelerator design is realized using Vitis high-level synthesis tool, using OpenCL, a language for heterogeneous computing, and evaluated on an Alveo U-50 FPGA card. The end-to-end ASR system has a latency of ∼120ms, which is suitable for real-time applications. The design demonstrates an average speed-up of 32× compared to an Intel Xeon E5-2640 CPU and 8.8× compared to NVIDIA GeForce RTX 3080 Ti Graphics card for a 32-bit floating point single precision model.