The performance of convolutional neural networks (CNN) depends heavily on their architectures. Transfer learning performance of a CNN relies quite strongly on selection of its trainable layers. Selecting the most effe...
详细信息
To support massive random access trials of devices with diverse QoS is a major challenge for massive machine-type communications. Space-air-ground integrated network(SAGIN) can be a promising solution for the congesti...
详细信息
Text-based emotion identification goes beyond simple sentiment analysis by capturing emotions in a more nuanced way, akin to shades of gray rather than just positive or negative sentiments. This paper details our expe...
详细信息
ISBN:
(数字)9798350361537
ISBN:
(纸本)9798350361544
Text-based emotion identification goes beyond simple sentiment analysis by capturing emotions in a more nuanced way, akin to shades of gray rather than just positive or negative sentiments. This paper details our experiments with emotion analysis on Bangla text. We collected a corpus of user comments from various social media groups discussing socioeconomic and political topics to identify six emotions: sadness, disgust, surprise, fear, anger, and joy. We evaluated the performance of four widely used machine learning algorithms—RF, DT, k-NN, and SVM—alongside three popular deep learning algorithms—CNNs, LSTM, and Transformer Learning—using TF-IDF feature extraction and word embedding techniques. The results showed that among the machine learning algorithms, DT, RF, k-NN, and SVM achieved accuracy scores of 82%, 84%, 73%, and 83%, respectively. In contrast, the deep learning models CNN and LSTM both achieved higher performance with an accuracy of 85% and 86% respectively. These findings highlight the effectiveness of traditional ML and DL approaches in detecting emotions from Bangla social media texts, indicating significant potential for further advancements in this area.
Few-Shot Learning (FSL) is a sub-area of machine learning which mainly deals with data where there is a scarcity of training supervised samples. Few shot learning (FSL) more closely resembles the human brain in compar...
Few-Shot Learning (FSL) is a sub-area of machine learning which mainly deals with data where there is a scarcity of training supervised samples. Few shot learning (FSL) more closely resembles the human brain in comparing new concepts to others based on prior experience rather than identifying it exactly. FSL aims to generalize the model across the tasks (in meta learning) opposed to the classical supervised learning which generalizes across the data points. In general the FSL models may suffer from underfitting because of scarcity of supervised samples and at the same time it causes overfitting as it is likely to memorize task specific features of the training set. This work aims to reduce such problems and is presented as a metric based model ”Few Shot Learning with Feature Pairing and Mean Discrepancy” (FL-FPMD). As the title suggests, feature pairing is one among various data augmentations. It is observed that flip augmentation is more suitable in the context of pairing the features within the given task. Memorizing task specific features is reduced by incorporating the discrepancy of mean distributions of the query and the support embedding in the loss function. The training and the evaluation is performed at the miniImageNet dataset and the results indicate that the proposed model outperforms the state-of-the-art models of similar complexity.
This paper describes a novel technique to improving Large Language Models (LLMs) for document analysis that employs knowledge graphs and retrieval-augmented generation (RAG). We are working on constructing a chatbot s...
详细信息
ISBN:
(数字)9798331517953
ISBN:
(纸本)9798331517960
This paper describes a novel technique to improving Large Language Models (LLMs) for document analysis that employs knowledge graphs and retrieval-augmented generation (RAG). We are working on constructing a chatbot system that can handle and analyze large documents from a variety of fields. Our approach addresses basic LLM issues including context maintenance and hallucination avoidance. The system combines document chunking, vector embedding, and similarity search with graph-based knowledge representation. Users can upload large papers and answer questions accurately. We show that integrating standard information retrieval approaches with graph-based storage and LLM capabilities improves context awareness and response accuracy across a wide range of document genres. This strategy is especially promising for complicated publications such as financial reports.
Natural Language Processing (NLP) has entered a new era with the advent of pre-trained language models, paving the way for constructing robust language models. Pretrained transformer-based models such as GPT-2 have be...
Natural Language Processing (NLP) has entered a new era with the advent of pre-trained language models, paving the way for constructing robust language models. Pretrained transformer-based models such as GPT-2 have become prevalent due to their cutting-edge efficiency. However, these approaches rely heavily on resource-intensive languages, forcing other languages to adopt multilingual frameworks (mGPT). The mGPT model could perform better for low-resource languages such as Bangla because the model has been trained on a diverse dataset spanning multiple languages. Recent studies show that the language-specific GPT model outperforms the multilingual mGPT model. In this research, we have proposed a pretrained monolingual GPT model called BanglaGPT using the objective of causal language modeling (CLM). Due to the lack of available large datasets for NLP tasks in Bangla, we have created a Bangla language model dataset called BanglaCLM using a 26.24 GB Bangla corpus scraped from several public websites. We have used a subword-based tokenization algorithm named Byte-Pair Encoding (BPE) for Bangla and finally trained the Bangla-GPT2 model from scratch using the BanglaCLM dataset. Our pretrained BanglaGPT provides state-of-the-art performance for Bangla text generation with a perplexity score of 2.86 and a loss score of 0.45 on the test set.
Resource virtualization is a promising technique that has been increasingly deployed in industrial automation systems to support multiple time-critical applications sharing the same physical resources. Extensive studi...
Resource virtualization is a promising technique that has been increasingly deployed in industrial automation systems to support multiple time-critical applications sharing the same physical resources. Extensive studies have been reported on how to perform real-time virtualization on computing resources. However, when applying virtualization techniques on network resources (especially for real-time wireless networks), node dependency among applications, wireless channel contention and stringent end-to-end timing requirements of the real-time flows in the network pose severe challenges. To address this problem, this paper formulates the network virtualization problem for multi-hop multi-channel real-time wireless networks (RTWNs). We first present a Satisfiability Modulo Theory (SMT)-based exact solution to capture the constraints posted by each application's resource interfaces and node dependency graphs. A novel supply graph (SG)-based partitioning framework, SGP, is then proposed to determine the resource partitions for individual applications. SGP uses supply graph to maintain compliance with the regularity constraints while efficiently allocating resources. Experimental results from both a real-world testbed and extensive simulations show that SGP can achieve comparable success ratio with the SMT-based exact solution but reduce the computational overhead significantly.
Smartphones are increasingly vital to people on a daily basis. Telephones are utilized in all aspects of life, ranging from personal to professional, due to technological advancements. It serves a function beyond maki...
详细信息
ISBN:
(数字)9798350306446
ISBN:
(纸本)9798350306453
Smartphones are increasingly vital to people on a daily basis. Telephones are utilized in all aspects of life, ranging from personal to professional, due to technological advancements. It serves a function beyond making phone calls. It enables internet connectivity and email reading. when not using the computer. The characteristics of a mobile phone are a crucial consideration when buying *** overall objective of this research is to find the best way to apply machine learning to estimate the retail pricing of smartphones based on their individual specs. Individuals who frequently use their phone are more attentive to selecting features. When purchasing a cell phone, a comparison is done based on the price-performance ratio. Phone features are regarded as performance. This research aims to forecast if mobile phones with certain features are considered economical or *** work is capable of being utilized in various marketing and business contexts to assist in making informed purchasing decisions by maximizing features while minimizing costs.
In this study, we demonstrate a graphene-based portable gas sensor with an integrated fused silica micro chamber for direct on-chip sample injection to detect a low concentration (-10 ppm) of acetone from a soil sampl...
详细信息
The primary purpose of the software industry is to provide high-quality software. Software system failure is caused by faulty software components. The goal of reliable software is to reduce the amount of software prog...
详细信息
暂无评论