In-context learning is a remarkable capability of transformers, referring to their ability to adapt to specific tasks based on a short history or context. Previous research has found that task-specific information is locally encoded within models, though their emergence and functionality remain unclear due to opaque pre-training processes. In this work, we investigate the formation of task vectors in a controlled setting, using models trained from scratch on synthetic datasets. Our findings confirm that task vectors naturally emerge under certain conditions, but the tasks may be relatively weakly and/or non-locally encoded within the model. To promote strong task vectors encoded at a prescribed location within the model, we propose an auxiliary training mechanism based on a task vector prompting loss (TVP-loss). This method eliminates the need to search for task-correlated encodings within the trained model and demonstrably improves robustness and generalization.

Voice biometric tasks, such as age estimation require modeling the often complex relationship between voice features and the biometric variable. While deep learning models can handle such complexity, they typically require large amounts of accurately labeled data to perform well. Such data are often scarce for biometric tasks such as voice-based age prediction. On the other hand, simpler models like linear regression can work with smaller datasets but often fail to generalize to the underlying non-linear patterns present in the data. In this paper we propose the Tessellated Linear Model (TLM), a piecewise linear approach that combines the simplicity of linear models with the capacity of non-linear functions. TLM tessellates the feature space into convex regions and fits a linear model within each region. We optimize the tessellation and the linear models using a hierarchical greedy partitioning. We evaluated TLM on the TIMIT dataset on the task of age prediction from voice, where it outperformed state-of-the-art deep learning models.

Accurate prediction of molecular activities is crucial for efficient drug discovery, yet remains challenging due to limited and noisy datasets. We introduce Similarity-Quantized Relative Learning (SQRL), a learning framework that reformulates molecular activity prediction as relative difference learning between structurally similar pairs of compounds. SQRL uses precomputed molecular similarities to enhance training of graph neural networks and other architectures, and significantly improves accuracy and generalization in low-data regimes common in drug discovery. We demonstrate its broad applicability and real-world potential through benchmarking on public datasets as well as proprietary industry data. Our findings demonstrate that leveraging similarity-aware relative differences provides an effective paradigm for molecular activity prediction.

This paper proposes the ViT Token Constraint and Multi-scale Memory bank (TCMM) method to address the patch noises and feature inconsistency in unsupervised person re-identification works. Many excellent methods use ViT features to obtain pseudo labels and clustering prototypes, then train the model with contrastive learning. However, ViT processes images by performing patch embedding, which inevitably introduces noise in patches and may compromise the performance of the re-identification model. On the other hand, previous memory bank based contrastive methods may lead data inconsistency due to the limitation of batch size. Furthermore, existing pseudo label methods often discard outlier samples that are difficult to cluster. It sacrifices the potential value of outlier samples, leading to limited model diversity and robustness. This paper introduces the ViT Token Constraint to mitigate the damage caused by patch noises to the ViT architecture. The proposed Multi-scale Memory enhances the exploration of outlier samples and maintains feature consistency. Experimental results demonstrate that our system achieves state-of-the-art performance on common benchmarks. The project is available at \href{https://github.com/andy412510/TCMM}{https://github.com/andy412510/TCMM}.

Ensemble Kalman inversion (EKI) is a derivative-free, particle-based optimization method for solving inverse problems. It can be shown that EKI approximates a gradient flow, which allows the application of methods for accelerating gradient descent. Here, we show that Nesterov acceleration is effective in speeding up the reduction of the EKI cost function on a variety of inverse problems. We also implement Nesterov acceleration for two EKI variants, unscented Kalman inversion and ensemble transform Kalman inversion. Our specific implementation takes the form of a particle-level nudge that is demonstrably simple to couple in a black-box fashion with any existing EKI variant algorithms, comes with no additional computational expense, and with no additional tuning hyperparameters. This work shows a pathway for future research to translate advances in gradient-based optimization into advances in gradient-free Kalman optimization.

A key challenge in Deep Reinforcement Learning is sample efficiency, especially in real-world applications where collecting environment interactions is expensive or risky. Recent off-policy algorithms improve sample efficiency by increasing the Update-To-Data (UTD) ratio and performing more gradient updates per environment interaction. While this improves sample efficiency, it significantly increases computational cost due to the higher number of gradient updates required. In this paper we propose a sample-efficient method to improve computational efficiency by separating training into distinct learning phases in order to exploit gradient updates more effectively. Our approach builds on top of the Dropout Q-Functions (DroQ) algorithm and alternates between an online, low UTD ratio training phase, and an offline stabilization phase. During the stabilization phase, we fine-tune the Q-functions without collecting new environment interactions. This process improves the effectiveness of the replay buffer and reduces computational overhead. Our experimental results on continuous control problems show that our method achieves results comparable to state-of-the-art, high UTD ratio algorithms while requiring 56\% fewer gradient updates and 50\% less training time than DroQ. Our approach offers an effective and computationally economical solution while maintaining the same sample efficiency as the more costly, high UTD ratio state-of-the-art.

Event prediction tasks often handle spatio-temporal data distributed in a large spatial area. Different regions in the area exhibit different characteristics while having latent correlations. This spatial heterogeneity and correlations greatly affect the spatio-temporal distributions of event occurrences, which has not been addressed by state-of-the-art models. Learning spatial dependencies of events in a continuous space is challenging due to its fine granularity and a lack of prior knowledge. In this work, we propose a novel Graph Spatio-Temporal Point Process (GSTPP) model for fine-grained event prediction. It adopts an encoder-decoder architecture that jointly models the state dynamics of spatially localized regions using neural Ordinary Differential Equations (ODEs). The state evolution is built on the foundation of a novel Self-Adaptive Anchor Graph (SAAG) that captures spatial dependencies. By adaptively localizing the anchor nodes in the space and jointly constructing the correlation edges between them, the SAAG enhances the model's ability of learning complex spatial event patterns. The proposed GSTPP model greatly improves the accuracy of fine-grained event prediction. Extensive experimental results show that our method greatly improves the prediction accuracy over existing spatio-temporal event prediction approaches.

Large language models (LLMs) have spurred development in multiple industries. However, the growing number of their parameters brings substantial storage and computing burdens, making it essential to explore model compression techniques for parameter reduction and easier deployment. We propose SWSC, an LLM compression method based on the concept of Shared Weight for Similar Channel. It uses the K-Means clustering algorithm to cluster model weights channel-by-channel, generating clusters with highly similar vectors within each. A representative vector from each cluster is selected to approximately replace all vectors in the cluster, significantly reducing the number of model weight parameters. However, approximate restoration will inevitably cause damage to the performance of the model. To tackle this issue, we perform singular value decomposition on the weight error values before and after compression and retain the larger singular values and their corresponding singular vectors to compensate for the accuracy. The experimental results show that our method can effectively ensure the performance of the compressed LLM even under low-precision conditions.

With the growing complexity of Cyber-Physical Systems (CPS) and the integration of Internet of Things (IoT), the use of sensors for online monitoring generates large volume of multivariate time series (MTS) data. Consequently, the need for robust anomaly diagnosis in MTS is paramount to maintaining system reliability and safety. While significant advancements have been made in anomaly detection, localization remains a largely underexplored area, though crucial for intelligent decision-making. This paper introduces a novel transformer-based model for unsupervised anomaly diagnosis in MTS, with a focus on improving localization performance, through an in-depth analysis of the self-attention mechanism's learning behavior under both normal and anomalous conditions. We formulate the anomaly localization problem as a three-stage process: time-step, window, and segment-based. This leads to the development of the Space-Time Anomaly Score (STAS), a new metric inspired by the connection between transformer latent representations and space-time statistical models. STAS is designed to capture individual anomaly behaviors and inter-series dependencies, delivering enhanced localization performance. Additionally, the Statistical Feature Anomaly Score (SFAS) complements STAS by analyzing statistical features around anomalies, with their combination helping to reduce false alarms. Experiments on real world and synthetic datasets illustrate the model's superiority over state-of-the-art methods in both detection and localization tasks.

The contextual bandit problem, which is a type of reinforcement learning tasks, provides an effective framework for solving challenges in recommendation systems, such as satisfying real-time requirements, enabling personalization, addressing cold-start problems. However, contextual bandit algorithms face challenges since they need to handle large state-action spaces sequentially. These challenges include the high costs for learning and balancing exploration and exploitation, as well as large variations in performance that depend on the domain of application. To address these challenges, Tsuboya et~al. proposed the Regional Linear Risk-sensitive Satisficing (RegLinRS) algorithm. RegLinRS switches between exploration and exploitation based on how well the agent has achieved the target. However, the reward expectations in RegLinRS are linearly approximated based on features, which limits its applicability when the relationship between features and reward expectations is non-linear. To handle more complex environments, we proposed Neural Risk-sensitive Satisficing (NeuralRS), which incorporates neural networks into RegLinRS, and demonstrated its utility.