http://repository.iiitd.edu.in/xmlui/handle/123456789/1062 Sat, 11 Apr 2026 15:39:17 GMT 2026-04-11T15:39:17Z http://repository.iiitd.edu.in/xmlui/handle/123456789/1355 Extreme abstractive text summarization Simran; Chakraborty, Tanmoy (Advisor); Akhtar, Md. Shad (Advisor) Extreme Abstractive Summarization of long scientific papers requires domain knowledge and a concise summary maintaining faithfulness to the source and covering novel aspects presented in the paper. Human annotations are indeed expensive for the task, so we propose ExGrapf2, a novel encoder architecture that uses fractality, FFT, and Graph Convolution Network as its strong foundation to address the challenge. We observed that when the model is presented with different views of the source, it extracts more information from the same amount of data. ExGrapf2 successfully accomplishes the objective and beats the state-of-the-art models on SciTLDR dataset without any data augmentation. We also used the contrastive loss to enhance the performance further. The novelty is not only for the modules but also for how we fuse them. Thu, 01 Dec 2022 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/1355 2022-12-01T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/1353 Sarcasm explanation in multimodal dialogues Vaibhav, Shashwat; Akhtar, Md. Shad (Advisor) Sarcasm is a means to convey ridicule or contempt. There has been a plethora of work in sentiment analysis, of which sarcasm is one of the most challenging tasks due to the incongruity between the surface level and the intended meaning of the sarcastic remark. In a dialogue setting, comprehending an ironic utterance is challenging, especially when the context is unclear. The classical studies primarily dealt with sarcasm detection tasks which considered the textual modality as the prime one. These works did well in sarcasm detection but failed to provide any explanation behind the elicited sarcasm, resulting from the lack of understanding and comprehension of a sarcastic utterance. The hidden semantic meaning of a sarcastic utterance is difficult to grasp without complete contextual clues, such as acoustic and visual signals. To this end, we explore the task of multimodal sarcasm explanation in dialogues, which deals with generating a natural language explanation for any given sarcastic instance. In this work, we are proposing the Explanans (What to explain) and Explanandum (what does the explaining) and follow various psychological theories to explain the satirical discourse. We show quantitative and qualitative analysis of the proposed model and discuss a possible future venture which can involve the incorporation of cognitive features as well. Thu, 01 Dec 2022 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/1353 2022-12-01T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/1350 Microservice-based in-network AES solution for FPGA NICs Hussain, Lasani Data centers demand high throughput (100 to 400 Gbps) and sub-millisecond latency. The performance of data center applications heavily depends on the efficiency of the underlying TCP stack. Despite several optimizations, such as kernel bypass and zero copying, TCP processing consumes up to 60% of the entire CPU cycles for short-lived connections. Modern data centers are pushing the TCP processing to programmable data plane hardware (smart NICs) to improve performance and save CPU cycles. However, the user space application processes the transport layer security (TLS) functions, negating the benefits of TCP offload. Some research proposes offloading TLS state and connection and management but ignores the processing of compute-intensive TLS crypto algorithms. We aim to offer in-network crypto primitives that TLS offload solutions can incorporate. Our goal is to design an in-network crypto framework that promises high-speed, low latency, scalability, dynamic reconfiguration, and low-power by leveraging FPGA-based network hardware. This thesis presents an FPGA-based AES offload solution aiming to satisfy the required objectives. Thu, 01 Dec 2022 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/1350 2022-12-01T00:00:00Z http://repository.iiitd.edu.in/xmlui/handle/123456789/1347 Knowledge graph assisted extreme summarization of scientific documents in hyperbolic space Mukherjee, Asmita; Chakraborty, Tanmoy (Advisor); Akhtar, Md. Shad (Advisor) A research paper is a document that presents an original work and introduces new concepts and makes interconnections between them via arguments and statements. Extreme Summarization involves high compression of the information content of the document and representing it in a concise, meaningful format. Hence extreme summarization of scientific documents entails representation of the key concepts as presented in the concerned document in a concise and coherent manner.The challenge of this task is to capture the essential concepts as presented in the entire paper. In this paper we handle the problem of abstractive extreme summarization of scientific documents.The technique of abstractive summarization would allow us to concisely represent the information present in the entire scientific research paper by generating a new sentence rather than being constrained to having to pick sentences that is already present in the document. Since no single sentence in the document can capture the entire information presented in the document. In order to do effective extreme summarization of scientific documents,we propose Knowledge graph assisted hyperbolic BART(KAHB) , a knowledge graph assisted sequence to sequence architecture while transforming the intermediate embeddings into hyperbolic space. A knowledge graph helps to capture the interconnection between the concepts as presented in the paper, which a plain sequence to sequence model fails to do. Transforming the embeddings to an hyperbolic space helps to capture the inherent hierarchical relationships present in the document .Applying the above methods , we are able to achieve improvements in the performance of a standard sequence to sequence mode. Thu, 01 Dec 2022 00:00:00 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/1347 2022-12-01T00:00:00Z