Traditional Mixture-of-Experts (MoE) networks benefit from utilizing multiple smaller expert models as opposed to a single large network. However, these experts typically operate independently, leaving a question open about whether interconnecting these models could enhance the performance of MoE networks. In response, we introduce GRAPHMOE, a novel method aimed at augmenting the cognitive depth of language models via a self-rethinking mechanism constructed on Pseudo GraphMoE networks. GRAPHMOE employs a recurrent routing strategy to simulate iterative thinking steps, thereby facilitating the flow of information among expert nodes. We implement the GRAPHMOE architecture using Low-Rank Adaptation techniques (LoRA) and conduct extensive experiments on various benchmark datasets. The experimental results reveal that GRAPHMOE outperforms other LoRA based models, achieving state-of-the-art (SOTA) performance. Additionally, this study explores a novel recurrent routing strategy that may inspire further advancements in enhancing the reasoning capabilities of language models.

Model-based offline Reinforcement Learning (RL) allows agents to fully utilise pre-collected datasets without requiring additional or unethical explorations. However, applying model-based offline RL to online systems presents challenges, primarily due to the highly suboptimal (noise-filled) and diverse nature of datasets generated by online systems. To tackle these issues, we introduce the Causal Prompting Reinforcement Learning (CPRL) framework, designed for highly suboptimal and resource-constrained online scenarios. The initial phase of CPRL involves the introduction of the Hidden-Parameter Block Causal Prompting Dynamic (Hip-BCPD) to model environmental dynamics. This approach utilises invariant causal prompts and aligns hidden parameters to generalise to new and diverse online users. In the subsequent phase, a single policy is trained to address multiple tasks through the amalgamation of reusable skills, circumventing the need for training from scratch. Experiments conducted across datasets with varying levels of noise, including simulation-based and real-world offline datasets from the Dnurse APP, demonstrate that our proposed method can make robust decisions in out-of-distribution and noisy environments, outperforming contemporary algorithms. Additionally, we separately verify the contributions of Hip-BCPDs and the skill-reuse strategy to the robustness of performance. We further analyse the visualised structure of Hip-BCPD and the interpretability of sub-skills. We released our source code and the first ever real-world medical dataset for precise medical decision-making tasks.

Type 1 diabetes mellitus (T1D) is characterized by insulin deficiency and blood glucose (BG) control issues. The state-of-the-art solution for continuous BG control is reinforcement learning (RL), where an agent can dynamically adjust exogenous insulin doses in time to maintain BG levels within the target range. However, due to the lack of action guidance, the agent often needs to learn from randomized trials to understand misleading correlations between exogenous insulin doses and BG levels, which can lead to instability and unsafety. To address these challenges, we propose an introspective RL based on Counterfactual Invertible Neural Networks (CINN). We use the pre-trained CINN as a frozen introspective block of the RL agent, which integrates forward prediction and counterfactual inference to guide the policy updates, promoting more stable and safer BG control. Constructed based on interpretable causal order, CINN employs bidirectional encoders with affine coupling layers to ensure invertibility while using orthogonal weight normalization to enhance the trainability, thereby ensuring the bidirectional differentiability of network parameters. We experimentally validate the accuracy and generalization ability of the pre-trained CINN in BG prediction and counterfactual inference for action. Furthermore, our experimental results highlight the effectiveness of pre-trained CINN in guiding RL policy updates for more accurate and safer BG control.

The representation of knowledge based on first-order logic captures the richness of natural language and supports multiple probabilistic inference models. Although symbolic representation enables quantitative reasoning with statistical probability, it is difficult to utilize with machine learning models as they perform numerical operations. In contrast, knowledge embedding (i.e., high-dimensional and continuous vectors) is a feasible approach to complex reasoning that can not only retain the semantic information of knowledge but also establish the quantifiable relationship among them. In this paper, we propose recursive neural knowledge network (RNKN), which combines medical knowledge based on first-order logic with recursive neural network for multi-disease diagnosis. After RNKN is efficiently trained from manually annotated Chinese Electronic Medical Records (CEMRs), diagnosis-oriented knowledge embeddings and weight matrixes are learned. Experimental results verify that the diagnostic accuracy of RNKN is superior to that of some classical machine learning models and Markov logic network (MLN). The results also demonstrate that the more explicit the evidence extracted from CEMRs is, the better is the performance achieved. RNKN gradually exhibits the interpretation of knowledge embeddings as the number of training epochs increases.