8 papers
CipherGenome: Homomorphic Inference for Genomic Mixture-of-Experts
Guang Yang, Fengchen Liu
Genome foundation models are growing into sparse mixture-of-experts (MoE) networks whose expert weights no longer fit on the machines that hold the sequences, yet sending a private…
GenomeOcean Anywhere: Private WebGPU Inference for Genome MoEs
Guang Yang, Fengchen Liu
Genome foundation models are most useful where sequences are generated, yet the largest models need datacenter accelerators and a place to send private DNA. We ask whether a 15-bil…
GenoTrace: Inheritable Watermarks for Genome Foundation Model Distillation
Guang Yang, Fengchen Liu
Can a genome model retain a detectable record of the synthetic sequences used to train it? We study watermark inheritance through distillation with GenoTrace, a codon-aware extensi…
Asymmetric Phase Coding Video Watermarking
Guang Yang, Fengchen Liu
Existing video watermarking systems are symmetric: the party that can verify a mark holds the extractor weights or generator secret and can therefore also embed one. Benchmarks con…
Growing a Stand, Not a Tree: Joint Canopy Generation Reproduces Crown Shyness
Guang Yang, Fengchen Liu
In closed forests, neighboring tree crowns often stop short of touching, leaving a network of narrow gaps known as crown shyness. The pattern belongs to the stand rather than to an…
How China-Origin Vision-Language Models Move from Refusal to Reframing in State Alignment
Guang Yang, Fengchen Liu, Alex Wang +2
State-aligned distortion has been documented in China-origin text-based large language models (LLMs), but whether, and in what form, it arises in multimodal systems has not been sy…
GLID: Gated Local Intrinsic Dimension Repairs the Blind Spots of Face-Forgery Detectors
Guang Yang, Fengchen Liu
Fine-tuned foundation-model detectors dominate face-forgery benchmarks, yet they stay blind to generator families absent from training. We present GLID, a detector that repairs thi…
Asking Back: Interaction-Layer Antidistillation Watermarks
Guang Yang, Amir Ghasemian, Fengchen Liu +3
Detecting unauthorized knowledge distillation from a deployed LLM API is hard because the defender controls neither the attacker's training pipeline nor the next-token logits. Exis…