Our open source codes

our developed open source tools are available to download in github: https://github.com/ASU-ESIC-FAN-Lab

Efficient AI
1. Deep Ternary Neural Network: Optimizing Deep Convolution Neural Network with Ternarized Weights and High Accuracy
2. Run-time dynamic network training: A Progressive Sub-network Searching Framework for Dynamic Inference
3. Rep-Net: Efficient On-Device Learning via Feature Reprogramming
4. KSM: Fast Multiple Task Adaption via Kernel-wise Soft Mask Learning
5. Get More at Once: Alternating Sparse Training with Gradient Correction for developing run-time dynamic neural network
6. LP-3DGS: Learning to Prune 3D Gaussian Splatting
7. Speedy Mast3r: a post-training optimization framework to improve speed of MASt3R based image matching
8. LoRAFusion: A Crossbar-aware Multi-task Adaption Framework via Efficient Fusion of Pretrained LoRA Modules
9. SafeguardGS: 3D Gaussian Primitive Pruning While Avoiding Catastrophic Scene Destruction
AI Security
1. Adversarial Example Defense through Parametric Noise Injection: to improve neural network robustness against adversarial input
2. Bit-Flip Attack: Crushing Neural Network with Progressive Bit Search: It introduces a Bit-Flip Attack (BFA) algorithm which search and identify the vulnerable bits within a quantized deep neural network
3. T-BFA: Targeted Bit-Flip Adversarial Weight Attack
4. TBT: Targeted Neural Network Attack with Bit Trojan: insert a Trojan through adversarial bit-flip attack
5. Defending and Harnessing the Bit-Flip based Adversarial Weight Attack
6. DeepSteal: Advanced Model Extractions Leveraging Efficient Weight Stealing in Memories
7. Deep-Dup: An Adversarial Weight Duplication Attack Framework to Crush Deep Neural Network in Multi-Tenant FPGA
8. Phantom: Privacy-Preserving Deep Neural Network Model Obfuscation in Heterogeneous TEE and GPU System
In Memory Computing
1. PIMA-SIM: cross-layer memory device to processing-in-memory application evaluation framework
Neuromorphic Computing
1. PytorX: End-to-End ReRAM Crossbar Non-ideal Effect Adaption for Neural Network Mapping