Conversational Information Seeking (CIS) requires systems to maintain context across turns, resolve referential ambiguities, and adapt to evolving user intent [6]. Beyond contextual understanding, truly effective CIS demands personalisation: adapting retrieval and responses to each user's unique background and preferences [2]. However, two fundamental limitations constrain current systems. First, existing conversational search pipelines apply a single static workflow to every query, lacking the adaptability to decide when personalisation helps and when it degrades retrieval precision. Second, personalisation in standard benchmarks typically relies on clean, hand-authored oracle user profiles, ignoring the reality that real-world user context is buried within noisy, long-term interaction logs. This research addresses both limitations by proposing a novel modular, agentic framework, realised through a suite of specialised agents that are coordinated by a dynamic orchestration mechanism.

Conversational search systems help users satisfy complex information needs through natural language interactions, yet incorporating user preferences into ranking remains challenging. Existing rewrite-then-rerank pipelines capture topical relevance but struggle with fine-grained constraints such as negative preferences or formatting requirements. Instruction-following retrieval approaches are promising for enforcing such constraints, yet their use in personalised conversational search remains underexplored, since ''instructions'' within this context are implicit and situated within user history and profiles, rather than being explicitly stated. We show that instruction-following models can assist with complex queries but introduce noise and latency on simpler keyword queries. To address this issue, we propose Adaptive Personalised Reranking (APR), a framework that routes queries based on intent. APR uses efficient similarity-based reranking for simple queries and dynamically generates tailored instructions to guide an instruction-following reranker for constraint-heavy contexts. Oracle analysis on TREC iKAT 2023 and 2024 shows that instruction-following provides a ''rescue'' potential for hard queries. We also show that APR trained with synthetic data performs competitively against strong baselines such as MonoT5 while offering promising new research avenues.

Agents powered by large language models (LLMs) have demonstrated strong capabilities in a wide range of complex, real-world applications. However, LLM agents with a compromised memory bank may easily produce harmful outputs when the past records retrieved for demonstration are malicious. In this paper, we propose a novel Memory INJection Attack, MINJA, without assuming that the attacker can directly modify the memory bank of the agent. The attacker injects malicious records into the memory bank by only **interacting with the agent via queries and output observations**. These malicious records are designed to elicit a sequence of malicious reasoning steps corresponding to a different target query during the agent's execution of the victim user's query. Specifically, we introduce a sequence of *bridging steps* to link victim queries to the malicious reasoning steps. During the memory injection, we propose an *indication prompt* that guides the agent to autonomously generate similar bridging steps, with a *progressive shortening strategy* that gradually removes the indication prompt, such that the malicious record will be easily retrieved when processing later victim queries. Our extensive experiments across diverse agents demonstrate the effectiveness of MINJA in compromising agent memory. With minimal requirements for execution, MINJA enables any user to influence agent memory, highlighting the risk.

With the rise of machine learning techniques, ensuring the fairness of decisions made by machine learning algorithms has become of great importance in critical applications. However, measuring fairness often requires full access to the model parameters, which compromises the confidentiality of the models. In this paper, we propose a solution using zero-knowledge proofs, which allows the model owner to convince the public that a machine learning model is fair while preserving the secrecy of the model. To circumvent the efficiency barrier of naively proving machine learning inferences in zero-knowledge, our key innovation is a new approach to measure fairness only with model parameters and some aggregated information of the input, but not on any specific dataset. To achieve this goal, we derive new bounds for the fairness of logistic regression and deep neural network models that are tighter and better reflecting the fairness compared to prior work. Moreover, we develop efficient zero-knowledge proof protocols for common computations involved in measuring fairness, including the spectral norm of matrices, maximum, absolute value, and fixed-point arithmetic. We have fully implemented our system, FairZK, that proves machine learning fairness in zero-knowledge. Experimental results show that Fairzk is significantly faster than the naive approach and an existing scheme that use zero-knowledge inferences as a subroutine. The prover time is improved by 3.1x-1789x depending on the size of the model and the dataset. FairZK can scale to a large model with 47 million parameters for the first time, and generates a proof for its fairness in 343 seconds. This is estimated to be 4 orders of magnitude faster than existing schemes, which only scale to small models with hundreds to thousands of parameters.

Massive amounts of uncertain demand response (DR) lead to a sudden increase in the data flow of power and thermal cyber-physical system (PTCPS), greatly increasing the blocking risk on critical communication paths of PTCPS. This article proposes a two-stage robust coordinated planning (TRCP) model with decision-dependent uncertainty (DDU), aiming to minimize the investment and improve communication security. First, a hybrid networking strategy, including wired and wireless components, is proposed, and the area of the base station that covers the power line carrier (PLC) line can be replaced by wireless components. Second, a scalable DDU set is built, which links the data flow and the planning decision variables, and the influence of the user’s willingness to participate in DR is considered. Meanwhile, the DDU set can be degenerated into a decision-independent uncertainty (DIU) set or a deterministic set by changing the model coefficients. Then, to solve the proposed model, a precise robust cuts method is presented with a detailed derivation. In this method, static robust cut constraints are adopted, whereby all information of the uncertainty set can be considered. Case studies show that the proposed model can effectively improve the economics and security of PTCPS communication path planning.

Reciprocal cooperation is prevalent in human society. Understanding human reciprocal cooperation in human-agent interaction can help design human-agent systems that promote cooperation and joint performance. Studies have found that people reciprocate cooperative behavior when interacting with computer agents in social dilemma games. However, few studies have investigated human reciprocal cooperation with agents in complex dynamic environments. This article examines the behavioral and neurological patterns of reciprocal cooperation in a hospital management microworld experiment. The participants (n = 30) work with both high- and low-cooperation computer agents to share resources to cope with dynamic demands. Participants' resource sharing behaviors were recorded and their prefrontal cortex (PFC) activation was measured using functional near-infrared spectroscopy (fNIRS) technology. Similar to previous studies conducted with participants in the United States, results demonstrate that participants in China showed reciprocal cooperation behaviors with the agents. Specifically, participants share more resources and achieve higher performance when working with a high-cooperation agent than with a low-cooperation agent. A high activation level is detected in the right dorsolateral PFC when working with a high-cooperation agent. Other PFC activation patterns imply that cooperation could be unnecessarily mentally taxing in certain situations. These findings suggest that human cooperativeness in human-agent systems can be calibrated by an agent's cooperation behavior. System designers should design for appropriate cooperativeness and avoid the inefficient use of system resources. Neurological measures could be a useful tool to investigate the mental process in human-agent cooperation.

Dynamic demands are the basic inputs of some existing dynamic assignment function. However, current data collection technologies do not directly support all the requirements of Dynamic Traffic Assignment (DTA) models. As a new traffic data collection technology, cell probe data could provide link flow, travel time and dynamic traffic demand at the same time. But these parameters are partial for a network.

We present two complementary methods for automatically improving mesh parameterizations and demonstrate that they provide a very desirable combination of efficiency and quality. First, we describe a new iterative method for constructing quasi-conformal parameterizations with free boundaries. We formulate the problem as fitting the coordinate gradients to two guidance vector fields of equal magnitude that are everywhere orthogonal. In only one linear step, our method efficiently generates parameterizations with natural boundaries from those with convex boundaries. If repeated until convergence, it produces the unique global minimizer of the Dirichlet energy. Next, we introduce a new non-linear optimization framework that can rapidly reduce interior distortion under a variety of metrics. By iteratively solving linear systems, our algorithm converges to a high quality, low distortion parameterization in very few iterations. The two components of our system are effective both in combination or when used independently.

The work carried out to explore the feasibility of reconstructing human constrained motion manipulation skills is reported. This is achieved by tracing and learning the manipulation performed by a human operator in a haptic rendered virtual environment The peg-in-hole insertion problem is used as a case study. In the developed system, position and contact force and torque as well as orientation data generated in the haptic rendered virtual environment combined with a priori knowledge about the task are used to identify and learn the skills in the newly demonstrated task. The data obtained from the virtual environment is classified into different cluster sets using a competitive fuzzy clustering algorithm called Competitive Agglomeration (CA). The CA algorithm starts with an over specified number of clusters which compete for feature points in the training procedure. Clusters with small cardinalities lose the competition and gradually vanish. The optimal number of clusters that win the competition is eventually determined. The clusters in the optimum cluster set are tuned using Locally Weighted Regression (LWR) to produce prediction models for robot trajectory performing the physical assembly based on the force/position information received from the rig. A background on the work and its significance is provided. The approach developed is explained and the results obtained so far are presented.

Resampling raw surface meshes is one of the most fundamental operations used by nearly all digital geometry processing systems. The vast majority of this work has focused on triangular remeshing, yet quadrilateral meshes are preferred for many surface PDE problems, especially fluid dynamics, and are best suited for defining Catmull-Clark subdivision surfaces. We describe a fundamentally new approach to the quadrangulation of manifold polygon meshes using Laplacian eigenfunctions, the natural harmonics of the surface. These surface functions distribute their extrema evenly across a mesh, which connect via gradient flow into a quadrangular base mesh. An iterative relaxation algorithm simultaneously refines this initial complex to produce a globally smooth parameterization of the surface. From this, we can construct a well-shaped quadrilateral mesh with very few extraordinary vertices. The quality of this mesh relies on the initial choice of eigenfunction, for which we describe algorithms and hueristics to efficiently and effectively select the harmonic most appropriate for the intended application.

The paper carried out to explore the feasibility of reconstructing human constrained motion manipulation skills is reported. This is achieved by tracing and learning the manipulation performed by a human operator in a haptic rendered virtual environment. The peg-in-hole insertion problem is used as a case study. In the developed system, force and position variables generated in the haptic rendered virtual environment combined with a priori knowledge about the task are used to identify and learn the skills in the newly demonstrated task. The data obtained from the virtual environment is classified into different cluster sets using fuzzy Gustafson-Kessel model (FGK). Principal component analysis (PCA) is applied to each cluster to reduce the dimension of the data. The clusters in the optimum cluster set are tuned using locally weighted regression (LWR) to produce prediction models for robot trajectory performing the physical assembly based on the force/position information received from the rig