User interactions on online travel platforms (OTPs) naturally follow a two-stage booking funnel, spanning hotel click from a multi-hotel listing page and booking conversion through room selection within the clicked hotel. Crucially, booking decisions are set-dependent: users select among multiple room types within the same hotel, where price changes of one option can shift demand to others. Two challenges arise in modeling such behavior: (i) the cascading booking-funnel behavior from hotel click to booking conversion, and (ii) intra-hotel substitution across room types. While accurate modeling of such behavior is essential for demand prediction and downstream optimization, existing methods typically ignore this structure by collapsing user behavior into a single stage and overlooking intra-hotel substitution effects, resulting in biased demand estimations and suboptimal decisions. In this paper, we propose BFSNet, a Booking Funnel and Substitution-aware neural network that decomposes booking demand into hotel click and booking conversion, enabling interpretable and stage-wise learning of user behavior. To capture the intra-hotel substitution effect, we introduce a mixing network-based substitution effect module that incorporates economic structural information by enforcing monotonic substitution relationships between the demand of a focal room and the prices of competing room types. To bridge prediction and downstream decision-making, we further develop a lightweight pricing procedure that distills the learned prediction model to efficiently evaluate candidate price vectors under operational constraints. Experiments on large-scale production data from a major online travel platform demonstrate consistent improvements in demand prediction accuracy and significant gains in downstream revenue.

Dynamic coreset selection is a promising approach for improving the training efficiency of deep neural networks by periodically selecting a small subset of the most representative or informative samples, thereby avoiding the need to train on the entire dataset. However, it remains inherently challenging due not only to the complex interdependencies among samples and the evolving nature of model training, but also to a critical *coreset representativeness degradation issue* identified and explored in-depth in this paper, that is, the representativeness or information content of the coreset degrades over time as training progresses. Therefore, we argue that, in addition to designing accurate selection rules, it is equally important to endow the algorithms with the ability to assess the quality of the current coreset. Such awareness enables timely re-selection, mitigating the risk of overfitting to stale subsets—a limitation often overlooked by existing methods. To this end, this paper proposes an **E**fficient **R**epresentativeness-**A**ware **C**oreset **S**election method for deep neural networks, a lightweight framework that enables dynamic tracking and maintenance of coreset quality during training. While the ideal criterion—gradient discrepancy between the coreset and the full dataset—is computationally prohibitive, we introduce a scalable surrogate based on the signal-to-noise ratio (SNR) of gradients within the coreset, which is the main technical contribution of this paper and is also supported by our theoretical analysis. Intuitively, a decline in SNR indicates overfitting to the subset and declining representativeness. Leveraging this observation, our method triggers coreset updates without requiring costly Hessian or full-batch gradient computations, maintaining minimal computational overhead. Experiments on multiple datasets confirm the effectiveness of our approach. Notably, compared with existing gradient-based dynamic coreset selection baselines, our method achieves up to a 5.4\% improvement in test accuracy across multiple datasets.

Sparse and low-rank matrix composite approximation has emerged as a promising paradigm for compressing large language models (LLMs), offering a more flexible pruning structure than conventional methods based solely on sparse matrices. The significant variation in weight redundancy across layers, along with the differing rank and sparsity structures of weight matrices, makes identifying the globally optimal pruning structure extremely challenging. Existing methods often depend on uniform or manually designed heuristic rules to allocate weight sparsity across layers, subsequently compressing each matrix using matrix approximation techniques. Given the above theoretical difficulty in global compression of LLMs and the limited computational and data resources available compared to the training phase, we argue that a collaboration between learning and matrix approximation is essential for effective compression. In this paper, we propose a novel LLM compression framework based on generalized bilevel optimization that naturally formulates an effective collaborative mechanism. Specifically, the outer loop frames the weight allocation task as a probabilistic optimization problem, enabling the automatic learning of both layer-wise sparsities and matrix-wise retained ranks, while the inner loop solves the corresponding sparsity and rank-constrained model compression problem via matrix approximation. Our main technical contributions include two key innovations for efficiently solving this bilevel optimization problem. First, we introduce a truncated Gaussian prior-based probabilistic parameterization integrated with a policy gradient estimator, which avoids expensive backpropagation and stabilizes the optimization process. Second, we design an adapted QR-based matrix approximation algorithm that significantly accelerates inner loop computations. Extensive experiments on Phi-3 and the LLama-2/3 family demonstrate the effectiveness of our method. Notably, it maintains over 95\% zero-shot accuracy under 50\% sparsity and achieves up to 2× inference speedup.

Network virtualization has emerged as a significant trend to solve the issues caused by ossification of traditional network. Under the circumstance of network virtualization, substrate network and virtual network are inextricably interdepending each other. The substrate network serves many virtual networks. Substrate network faults may lead to different virtual network faults. A service''s failure may introduce additional influence on other services. Therefore, it has become a big challenge to predict when and where a fault happens in the network. In this paper, we propose a fault diagnosis method by deep learning to predict the failure of virtual network. Our deep learning model enables the earlier failure prediction by the Long Short-Term Memory (LSTM) network, which discovers the long-term features of network history data. Simulation results show that the proposed method performs well on faults prediction.

Biopsy of zona pellucida is a necessary step prior to pre-implantation genetic diagnosis (PGD). Besides traversing zona pellucida by applying laser or acidified medium (e.g Tyrode's solution), mechanical means is another safe approach. Traditional mechanical zona cutting requires highly skilled embryologist. It is very difficult to make the process automatic due to its complexity. The process can be enhanced by introducing piezo-driven cutter, which makes the cutting process more precise and introduces less cell deformation. In this paper, the application of lateral oscillation of the piezo-driven microneedle is introduced. We believe that further understanding and implementation of the lateral vibrational piezo-driven microcutter can be beneficial for more accurate and controllable mechanical cell biopsy.

To fulfill the needs for accurate trajectory tracking with large displacement in a handheld instrument, a 3-DOF serial compliant mechanism is developed. The mechanism is compact with a total length less than 150 mm and a maximum diameter of 22 mm. Two flexures are developed using different rapid prototyping techniques: one 3-DOF flexural lever made of Vero-Gray by Polyjet and a 1-DOF translational flexure made of stainless steel by Direct Metal Laser Sintering (DMLS). Analytical and Finite Element (FE) models are developed for the proposed flexural mechanisms. Experiments are conducted on a prototype. To improve the tracking accuracy, the hysteretic nonlinearities of the system are modeled using Prandtl-Ishlinskii model. Inverse feedforward controller is implemented to linearize the relationship between input and output. The tracking errors are reduced while maintaining a fast response of the system. The total tracking errors are identified individually for each axis and then compensated. Tracking performances of the tool tip are evaluated experimentally with different inputs. The RMS tracking error of the proposed mechanism is lower than 1 µm in all axes, which is improved more than four times compared to the previous systems.