Large Language Model (LLM) Optimization

• Optimizing retrieval models for faster and more accurate augmented generation
• Applying Low-Rank Adaptation (LoRA) to fine-tune large-scale models efficiently.
• Parameter-Efficient Fine-Tuning (PEFT)
• Time-Series data Anomaly Detection Model

Generative Models using Deep Learning

• GAN-Based Models
  • Generative Adversarial Networks (GANs)
  • Conditional GAN (cGAN)
  • CycleGAN
  • StyleGAN / StyleGAN2
• Autoencoder-Based Models
  • Variational Autoencoder (VAE)
  • Vector Quantized VAE (VQ-VAE)
  • Denoising Autoencoder
  • Sparse Autoencoder
• Diffusion-Based Models
  • Denoising Diffusion Probabilistic Models (DDPMs)
  • Score-Based Generative Models
  • Vector Quantized Diffusion Models (VQ-DM)
  • Conditional Diffusion Models with Quantization (CDM-Q)
  • Discrete Diffusion Models (DDM)

Image Segmentation using Deep Learning

• Supervised Image Segmentation
  • U-Net
  • DeepLab (DeepLabV3, DeepLabV3+)
  • Mask R-CNN
  • PSPNet (Pyramid Scene Parsing Network)
  • SegNet
• Unsupervised Image Segmentation
  • Mean Shift Segmentation
  • Deep Embedded Clustering (DEC)
  • Segment Anything Model (SAM)

Image Anomaly Detection

• Out-of-Distribution Detection
• Defect Generation
• Feature-Embedding Based Methods
• Reconstruction Based Methods

Paper Reading and Research

• Code for Research
  • BraTs 2017 Brain Tumor Segmentation
• Paper Reading
  • A Survey of Methods for Brain Tumor Segmentation Based on MRI Images
  • Edge U-Net: Brain Tumor Segmentation Using MRI Based on Deep U-Net Model with Boundary Information
  • FU-net: Multi-class Image Segmentation Using Feedback Weighted U-net
  • Using a Generative Adversarial Network to Generate Synthetic MRI Images for Multi-class Automatic Segmentation of Brain Tumors
  • Edge-Boosted U-Net for 2D Medical Image Segmentation
  • A survey on efficient vision transformers: algorithms, techniques, and performance benchmarking
  • others in here