computer science

Assignment 5: Advanced Topics in Data Classification

TASK 1: PUBMED 20K RCT: DATASET FOR SEQUENTIAL SENTENCE CLASSIFICATION

OBJECTIVE:

In this task, you will work with the PubMed 20k RCT dataset, a resource for sequential sentence classification. Your goal is to develop a model that can classify sentences in the context of their position in abstracts of biomedical research papers.

DATASET:

Source: The PubMed 20k RCT dataset

Description: The dataset comprises abstracts from randomized controlled trials, with sentences sequentially labeled.

Access: You can find the dataset in the Files section under the folder Assignment5 resources.

TASK 2: MULTI-FIREARM AUDIO CLASSIFICATION USING DEEP LEARNING

OBJECTIVE:

Develop a deep learning model that can classify audio recordings of firearm discharges, recognizing different types of firearms.

DATASET:

Description: A collection of audio recordings featuring various firearms being discharged in different orientations.

BASELINE MODEL:

A sample code for a baseline model will be provided and explained in the lecture. Please pay attention to the coming lectures.

You can download the data from the following link Gun shot datasetLinks to an external site..

REQUIREMENTS:

Audio Data Preprocessing: Implement techniques for processing and feature extraction from audio data.

Model Development: Create a deep learning model ( LSTM, CNN ) that can classify the type of firearm and its orientation based on the audio recording.

Evaluation: Evaluate your model using accuracy, precision, recall, and F1-score.

Report: Write a comprehensive report documenting your approach, experiments, results, and any interesting findings or challenges.

Discussion and Conclusion: Discuss the results and provide insights into the behavior of the different models with this particular dataset.

SUBMISSION GUIDELINES:

You need to submit the file as follows (firstname_lastname.ipynb)

Extra Credit Assignment ( 4 points) : Summarizing Research Papers on Gunshot Audio ClassificationTask: Your assignment is to read and summarize five academic papers related to gunshot audio classification. Each summary should concisely capture the main research question, methodology, and key findings of the paper. You can find the papers in the file section under the folder assignment5.Guidelines:

Research Question: Briefly describe the primary question or problem the paper addresses.

Methodology: Outline the methods or techniques used in the research.

Key Findings: Summarize the main results and conclusions drawn by the authors.

Submission Format:

Provide a two-sentence summary for each paper.

Include the title and authors of each paper in your submission.

Objective: This task aims to enhance your understanding of the current state of research in audio classification of gunshots and to develop your skills in critical analysis and concise communication.