Soham Nandi

Computer Science Engineering, specializing in Artificial Intelligence and Machine learning

Soham Nandi is a technical writer based in Memari, India. His academic background is in Computer Science Engineering, specializing in Artificial Intelligence and Machine learning. He has extensive experience in Data Analytics, Machine Learning, and Python. He has worked on group projects that required the implementation of Computer Vision, Image Classification, and App Development.

Soham's passion for AI and Machine Learning propels him toward his goal of becoming a distinguished Machine Learning Engineer and leveraging his knowledge to help the less fortunate. His projects include the Rice grain classifier and Handwriting-based Personality Detector, and he is certified in Data Analytics and Data Science.

Drawing inspiration from writers ranging from Satyajit Ray to Mark Manson, he has his own science fiction stories and blog. He also creates content on personal development and mental health through his Instagram handle. Beyond the screen and code, Soham's world is filled with music. With the ability to play the guitar, ukulele, and drums, he rocks to the tunes of the metal and rock genres, using music as a creative outlet, planning to start a band of his own.

See less

Articles from Soham

Enhancing Geographic Diversity in Text-to-Image Models with c-VSG

Researchers introduced contextualized Vendi score guidance (c-VSG) to address geographic diversity limitations in text-to-image generative models. By integrating real-world exemplar images and leveraging the Vendi Score (VS), c-VSG significantly improved image diversity across geographically diverse datasets like GeoDE and DollarStreet.

1 Jul 2024

Innovative DRL Approach to Alleviate Traffic Congestion

Researchers in Nature unveiled a new method for traffic signal control using deep reinforcement learning (DRL) that addresses convergence and robustness issues. The PN_D3QN model, incorporating dueling networks, double Q-learning, priority sampling, and noise parameters, processed high-dimensional traffic data and achieved faster convergence.

28 Jun 2024

Novel DIG Metrics for Fairer Image Generation

Researchers have introduced Decomposed-DIG, a set of metrics to evaluate geographic biases in text-to-image generative models by separately assessing objects and backgrounds in generated images. The study reveals significant regional disparities, particularly in Africa, and proposes a new prompting strategy to improve background diversity.

27 Jun 2024

DRAGON: Advanced Dense Retrieval with Diverse Augmentation

Researchers at Meta have developed DRAGON, a dense retriever model that leverages diverse data augmentation to achieve state-of-the-art effectiveness in both supervised and zero-shot retrieval tasks without increasing model complexity.

26 Jun 2024

Tackling AI Hallucinations for Reliable Outputs

Hallucinations in large language models result from biases in training data and inherent model architecture issues, leading to factually incorrect or misleading outputs. Understanding the causes and implementing prevention strategies, such as enhancing training data quality and refining algorithms, is crucial. Detection and correction techniques, including advanced NLP tools and human oversight, are essential to ensure the responsible deployment of AI technologies in various applications.

25 Jun 2024

Dynamic Neural Codebooks Boost Vector Quantization

Researchers introduced QINCo, a novel vector quantization method that employs neural networks to dynamically generate codebooks, significantly improving data compression and vector search accuracy. Experimental results demonstrated QINCo's superiority over existing methods, achieving better nearest-neighbor search performance with more compact code sizes across multiple datasets.

25 Jun 2024

Enhancing Financial Sentiment Analysis with Lexicon-Based Prompts

Researchers introduced a novel method using domain-specific lexicons to refine pre-trained language models for financial sentiment analysis. This approach improved accuracy without requiring extensive labeled data, demonstrating superior performance over traditional domain adaptation techniques across various models like BERT, RoBERTa, Electra, and T5

24 Jun 2024

Enhancing NMR Spectroscopy for Large Biomolecules Using AI

Researchers introduced a novel AI-based approach to NMR spectroscopy, allowing for high-resolution analysis of large biomolecules without the need for deuteration. This method significantly enhances the study of proteins up to 360 kDa, providing cost-effective and comprehensive data.

21 Jun 2024

ORACLE: Enhancing Wildlife Surveillance with Drones

Researchers developed ORACLE, an advanced computer vision model utilizing YOLO architecture for automated bird detection and tracking from drone footage. Achieving a 91.89% mean average precision, ORACLE significantly enhances wildlife conservation by accurately identifying and monitoring avian species in dynamic environments.

20 Jun 2024

Control and Motion Planning of Fixed-wing UAVs Through Reinforcement Learning

Researchers demonstrated how reinforcement learning (RL) can improve guidance, navigation, and control (GNC) systems for unmanned aerial vehicles (UAVs), enhancing robustness and efficiency in tasks like dynamic target interception and waypoint tracking.

19 Jun 2024