Heming Zhang

Wenjia Zhang, Haoran Xu, Haoyi Niu et al. · tsinghua

Offline imitation learning (IL) is a powerful method to solve decision-making problems from expert demonstrations without reward labels. Existing offline IL methods suffer from severe performance degeneration under limited expert data. Including a learned dynamics model can potentially improve the state-action space coverage of expert data, however, it also faces challenging issues like model approximation/generalization errors and suboptimality of rollout data. In this paper, we propose the Discriminator-guided Model-based offline Imitation Learning (DMIL) framework, which introduces a discriminator to simultaneously distinguish the dynamics correctness and suboptimality of model rollout data against real expert demonstrations. DMIL adopts a novel cooperative-yet-adversarial learning strategy, which uses the discriminator to guide and couple the learning process of the policy and dynamics model, resulting in improved model performance and robustness. Our framework can also be extended to the case when demonstrations contain a large proportion of suboptimal data. Experimental results show that DMIL and its extension achieve superior performance and robustness compared to state-of-the-art offline IL methods under small datasets.

Heming Zhang, Di Huang, Wenyu Li et al.

In precision medicine, quantitative multi-omic features, topological context, and textual biological knowledge play vital roles in identifying disease-critical signaling pathways and targets. Existing pipelines capture only part of these-numerical omics ignore topological context, text-centric LLMs lack quantitative grounded reasoning, and graph-only models underuse node semantics and the generalization of LLMs-limiting mechanistic interpretability. Although Process Reward Models (PRMs) aim to guide reasoning in LLMs, they remain limited by unreliable intermediate evaluation, and vulnerability to reward hacking with computational cost. These gaps motivate integrating quantitative multi-omic signals, topological structure with node annotations, and literature-scale text via LLMs, using subgraph reasoning as the principle bridge linking numeric evidence, topological knowledge and language context. Therefore, we propose GALAX (Graph Augmented LAnguage model with eXplainability), an innovative framework that integrates pretrained Graph Neural Networks (GNNs) into Large Language Models (LLMs) via reinforcement guided by a Graph Process Reward Model (GPRM), which generates disease-relevant subgraphs in a step-wise manner initiated by an LLM and iteratively evaluated by a pretrained GNN, enabling process-level supervision without explicit intermediate reasoning annotations. As an application, we also introduced Target-QA, a benchmark combining CRISPR-identified targets, multi-omic profiles, and biomedical graph knowledge across diverse cancer cell lines, which enables GNN pretraining for supervising step-wise graph construction and supports long-context reasoning over text-numeric graphs (TNGs), providing a scalable and biologically grounded framework for explainable, reinforcement-guided subgraph reasoning toward reliable and interpretable target and pathway discovery in precision medicine.

Zehao Dong, Heming Zhang, Yixin Chen et al.

We investigate molecular mechanisms of resistant or sensitive response of cancer drug combination therapies in an inductive and interpretable manner. Though deep learning algorithms are widely used in the drug synergy prediction problem, it is still an open problem to formulate the prediction model with biological meaning to investigate the mysterious mechanisms of synergy (MoS) for the human-AI collaboration in healthcare systems. To address the challenges, we propose a deep graph neural network, IDSP (Interpretable Deep Signaling Pathways), to incorporate the gene-gene as well as gene-drug regulatory relationships in synergic drug combination predictions. IDSP automatically learns weights of edges based on the gene and drug node relations, i.e., signaling interactions, by a multi-layer perceptron (MLP) and aggregates information in an inductive manner. The proposed architecture generates interpretable drug synergy prediction by detecting important signaling interactions, and can be implemented when the underlying molecular mechanism encounters unseen genes or signaling pathways. We test IDWSP on signaling networks formulated by genes from 46 core cancer signaling pathways and drug combinations from NCI ALMANAC drug combination screening data. The experimental results demonstrated that 1) IDSP can learn from the underlying molecular mechanism to make prediction without additional drug chemical information while achieving highly comparable performance with current state-of-art methods; 2) IDSP show superior generality and flexibility to implement the synergy prediction task on both transductive tasks and inductive tasks. 3) IDSP can generate interpretable results by detecting different salient signaling patterns (i.e. MoS) for different cell lines.

Xuewen Yang, Heming Zhang, Di Jin et al.

Generating accurate descriptions for online fashion items is important not only for enhancing customers' shopping experiences, but also for the increase of online sales. Besides the need of correctly presenting the attributes of items, the expressions in an enchanting style could better attract customer interests. The goal of this work is to develop a novel learning framework for accurate and expressive fashion captioning. Different from popular work on image captioning, it is hard to identify and describe the rich attributes of fashion items. We seed the description of an item by first identifying its attributes, and introduce attribute-level semantic (ALS) reward and sentence-level semantic (SLS) reward as metrics to improve the quality of text descriptions. We further integrate the training of our model with maximum likelihood estimation (MLE), attribute embedding, and Reinforcement Learning (RL). To facilitate the learning, we build a new FAshion CAptioning Dataset (FACAD), which contains 993K images and 130K corresponding enchanting and diverse descriptions. Experiments on FACAD demonstrate the effectiveness of our model.

Heming Zhang, Xiaolong Wang, C. -C. Jay Kuo

Kinship verification aims to identify the kin relation between two given face images. It is a very challenging problem due to the lack of training data and facial similarity variations between kinship pairs. In this work, we build a novel appearance and shape based deep learning pipeline. First we adopt the knowledge learned from general face recognition network to learn general facial features. Afterwards, we learn kinship oriented appearance and shape features from kinship pairs and combine them for the final prediction. We have evaluated the model performance on a widely used popular benchmark and demonstrated the superiority over the state-of-the-art.

Heming Zhang, Xiaolong Wang, Jingwen Zhu et al.

Face detection is a widely studied problem over the past few decades. Recently, significant improvements have been achieved via the deep neural network, however, it is still challenging to directly apply these techniques to mobile devices for its limited computational power and memory. In this work, we present a proposal generation acceleration framework for real-time face detection. More specifically, we adopt a popular cascaded convolutional neural network (CNN) as the basis, then apply our acceleration approach on the basic framework to speed up the model inference time. We are motivated by the observation that the computation bottleneck of this framework arises from the proposal generation stage, where each level of the dense image pyramid has to go through the network. In this work, we reduce the number of image pyramid levels by utilizing both global and local facial characteristics (i.e., global face and facial parts). Experimental results on public benchmarks WIDER-face and FDDB demonstrate the satisfactory performance and faster speed compared to the state-of-the-arts. %the comparable accuracy to state-of-the-arts with faster speed.

Jie Zhang, Junting Zhang, Shalini Ghosh et al.

Lifelong learning, the problem of continual learning where tasks arrive in sequence, has been lately attracting more attention in the computer vision community. The aim of lifelong learning is to develop a system that can learn new tasks while maintaining the performance on the previously learned tasks. However, there are two obstacles for lifelong learning of deep neural networks: catastrophic forgetting and capacity limitation. To solve the above issues, inspired by the recent breakthroughs in automatically learning good neural network architectures, we develop a Multi-task based lifelong learning via nonexpansive AutoML framework termed Regularize, Expand and Compress (REC). REC is composed of three stages: 1) continually learns the sequential tasks without the learned tasks' data via a newly proposed multi-task weight consolidation (MWC) algorithm; 2) expands the network to help the lifelong learning with potentially improved model capability and performance by network-transformation based AutoML; 3) compresses the expanded model after learning every new task to maintain model efficiency and performance. The proposed MWC and REC algorithms achieve superior performance over other lifelong learning algorithms on four different datasets.

Junting Zhang, Jie Zhang, Shalini Ghosh et al.

Deep neural networks (DNNs) often suffer from "catastrophic forgetting" during incremental learning (IL) --- an abrupt degradation of performance on the original set of classes when the training objective is adapted to a newly added set of classes. Existing IL approaches tend to produce a model that is biased towards either the old classes or new classes, unless with the help of exemplars of the old data. To address this issue, we propose a class-incremental learning paradigm called Deep Model Consolidation (DMC), which works well even when the original training data is not available. The idea is to first train a separate model only for the new classes, and then combine the two individual models trained on data of two distinct set of classes (old classes and new classes) via a novel double distillation training objective. The two existing models are consolidated by exploiting publicly available unlabeled auxiliary data. This overcomes the potential difficulties due to the unavailability of original training data. Compared to the state-of-the-art techniques, DMC demonstrates significantly better performance in image classification (CIFAR-100 and CUB-200) and object detection (PASCAL VOC 2007) in the single-headed IL setting.

Heming Zhang, Shalini Ghosh, Larry Heck et al.

The key challenge of generative Visual Dialogue (VD) systems is to respond to human queries with informative answers in natural and contiguous conversation flow. Traditional Maximum Likelihood Estimation (MLE)-based methods only learn from positive responses but ignore the negative responses, and consequently tend to yield safe or generic responses. To address this issue, we propose a novel training scheme in conjunction with weighted likelihood estimation (WLE) method. Furthermore, an adaptive multi-modal reasoning module is designed, to accommodate various dialogue scenarios automatically and select relevant information accordingly. The experimental results on the VisDial benchmark demonstrate the superiority of our proposed algorithm over other state-of-the-art approaches, with an improvement of 5.81% on recall@10.