Zihao Liu

EAAI Journal 2026 Journal Article

Corrigendum to “Representation enhancement and hierarchical semantic alignment for text–video retrieval” [Eng. Appl. Artif. Intell., 165, Part B, (1 February 2026), 113471]

• Jianqin Wu
• Xiaoyu Wu
• Jiayao Qian
• Zihao Liu

EAAI Journal 2026 Journal Article

Representation enhancement and hierarchical semantic alignment for text–video retrieval

• Jianqin Wu
• Xiaoyu Wu
• Jiayao Qian
• Zihao Liu

Text–video retrieval aims to retrieve videos corresponding to a given text description from a large-scale video library or to retrieve texts that match a specific video from a vast text corpus. Existing cross-modal methods for text–video retrieval face significant challenges in establishing strong semantic connections between text and video, primarily due to the limited text descriptions and the high inter-video similarity within datasets, which undermines the model’s ability to effectively represent features. To address these challenges, we propose a novel text–video retrieval method termed REHSA based on Representation Enhancement and Hierarchical Semantic Alignment. Our approach involves generating video caption as knowledge embedding and incorporating this information with video frames in the embedding space to enhance video representation. Furthermore, we propose a hierarchical semantic alignment module that includes fine-grained semantic alignment at the patch-word and frame–word levels, clustering center semantic alignment, and global semantic alignment. This module not only focuses on multi-grained feature alignment but also forms progressive associations between different levels, aligning more closely with human cognitive processes for interpreting text and video. Lastly, we implement cross-modal feature interaction through a mask language modeling task during training, further enhancing text and video representations. Extensive experiments demonstrate that our method achieves impressive performance, with 48. 6%, 49. 8%, and 43. 2% R@1 scores on the widely used benchmarks MSRVTT, MSVD, and ActivityNet Captions, respectively, surpassing most existing methods.

AAAI Conference 2026 Conference Paper

Zo3T: Zero-Shot 3D-Aware Trajectory-Guided Image-to-Video Generation via Test-Time Training

• Ruicheng Zhang
• Jun Zhou
• Zunnan Xu
• Zihao Liu
• Jiehui Huang
• Mingyang Zhang
• Yu Sun
• Xiu Li

Trajectory-Guided image-to-video (I2V) generation aims to synthesize videos that adhere to user-specified motion instructions. Existing methods typically rely on computationally expensive fine-tuning on scarce annotated datasets. Although some zero-shot methods attempt to trajectory control in the latent space, they may yield unrealistic motion by neglecting 3D perspective and creating a misalignment between the manipulated latents and the network's noise predictions. To address these challenges, we introduce Zo3T, a novel zero-shot test-time-training framework for trajectory-guided generation with three core innovations: First, we incorporate a 3D-Aware Kinematic Projection, leveraging inferring scene depth to derive perspective-correct affine transformations for target regions. Second, we introduce Trajectory-Guided Test-Time LoRA, a mechanism that dynamically injects and optimizes ephemeral LoRA adapters into the denoising network alongside the latent state. Driven by a regional feature consistency loss, this co-adaptation effectively enforces motion constraints while allowing the pre-trained model to locally adapt its internal representations to the manipulated latent, thereby ensuring generative fidelity and on-manifold adherence. Finally, we develop Guidance Field Rectification, which refines the denoising evolutionary path by optimizing the conditional guidance field through a one-step lookahead strategy, ensuring efficient generative progression towards the target trajectory. Zo3T significantly enhances 3D realism and motion accuracy in trajectory-controlled I2V generation, demonstrating superior performance over existing training-based and zero-shot approaches.

AAAI Conference 2024 Conference Paper

Backdoor Attacks via Machine Unlearning

• Zihao Liu
• Tianhao Wang
• Mengdi Huai
• Chenglin Miao

As a new paradigm to erase data from a model and protect user privacy, machine unlearning has drawn significant attention. However, existing studies on machine unlearning mainly focus on its effectiveness and efficiency, neglecting the security challenges introduced by this technique. In this paper, we aim to bridge this gap and study the possibility of conducting malicious attacks leveraging machine unlearning. Specifically, we consider the backdoor attack via machine unlearning, where an attacker seeks to inject a backdoor in the unlearned model by submitting malicious unlearning requests, so that the prediction made by the unlearned model can be changed when a particular trigger presents. In our study, we propose two attack approaches. The first attack approach does not require the attacker to poison any training data of the model. The attacker can achieve the attack goal only by requesting to unlearn a small subset of his contributed training data. The second approach allows the attacker to poison a few training instances with a pre-defined trigger upfront, and then activate the attack via submitting a malicious unlearning request. Both attack approaches are proposed with the goal of maximizing the attack utility while ensuring attack stealthiness. The effectiveness of the proposed attacks is demonstrated with different machine unlearning algorithms as well as different models on different datasets.

NeurIPS Conference 2020 Conference Paper

ShiftAddNet: A Hardware-Inspired Deep Network

• Haoran You
• Xiaohan Chen
• Yongan Zhang
• Chaojian Li
• Sicheng Li
• Zihao Liu
• Zhangyang Wang
• Yingyan Lin

Multiplication (e. g. , convolution) is arguably a cornerstone of modern deep neural networks (DNNs). However, intensive multiplications cause expensive resource costs that challenge DNNs' deployment on resource-constrained edge devices, driving several attempts for multiplication-less deep networks. This paper presented ShiftAddNet, whose main inspiration is drawn from a common practice in energy-efficient hardware implementation, that is, multiplication can be instead performed with additions and logical bit-shifts. We leverage this idea to explicitly parameterize deep networks in this way, yielding a new type of deep network that involves only bit-shift and additive weight layers. This hardware-inspired ShiftAddNet immediately leads to both energy-efficient inference and training, without compromising the expressive capacity compared to standard DNNs. The two complementary operation types (bit-shift and add) additionally enable finer-grained control of the model's learning capacity, leading to more flexible trade-off between accuracy and (training) efficiency, as well as improved robustness to quantization and pruning. We conduct extensive experiments and ablation studies, all backed up by our FPGA-based ShiftAddNet implementation and energy measurements. Compared to existing DNNs or other multiplication-less models, ShiftAddNet aggressively reduces over 80% hardware-quantified energy cost of DNNs training and inference, while offering comparable or better accuracies. Codes and pre-trained models are available at https: //github. com/RICE-EIC/ShiftAddNet.