Temporal Point Processes (TPPs), especially Hawkes Process are commonly used for modeling asynchronous event sequences data such as financial transactions and user behaviors in social networks. Due to the strong fitting ability of neural networks, various neural Temporal Point Processes are proposed, among which the Neural Hawkes Processes based on self-attention such as Transformer Hawkes Process (THP) achieve distinct performance improvement. Although the THP has gained increasing studies, it still suffers from the {sequence prediction issue}, i.e., training on history sequences and inferencing about the future, which is a prevalent paradigm in realistic sequence analysis tasks. What's more, conventional THP and its variants simply adopt initial sinusoid embedding in transformers, which shows performance sensitivity to temporal change or noise in sequence data analysis by our empirical study. To deal with the problems, we propose a new Rotary Position Embedding-based THP (RoTHP) architecture in this paper. Notably, we show the translation invariance property and {sequence prediction flexibility} of our RoTHP induced by the {relative time embeddings} when coupled with Hawkes process theoretically. Furthermore, we demonstrate empirically that our RoTHP can be better generalized in sequence data scenarios with timestamp translations and in sequence prediction tasks.
In the high-stakes world of baseball, every nuance of a pitcher's mechanics holds the key to maximizing performance and minimizing runs. Traditional analysis methods often rely on pre-recorded offline numerical data, hindering their application in the dynamic environment of live games. Broadcast video analysis, while seemingly ideal, faces significant challenges due to factors like motion blur and low resolution. To address these challenges, we introduce PitcherNet, an end-to-end automated system that analyzes pitcher kinematics directly from live broadcast video, thereby extracting valuable pitch statistics including velocity, release point, pitch position, and release extension. This system leverages three key components: (1) Player tracking and identification by decoupling actions from player kinematics; (2) Distribution and depth-aware 3D human modeling; and (3) Kinematic-driven pitch statistics. Experimental validation demonstrates that PitcherNet achieves robust analysis results with 96.82% accuracy in pitcher tracklet identification, reduced joint position error by 1.8mm and superior analytics compared to baseline methods. By enabling performance-critical kinematic analysis from broadcast video, PitcherNet paves the way for the future of baseball analytics by optimizing pitching strategies, preventing injuries, and unlocking a deeper understanding of pitcher mechanics, forever transforming the game.
The employment of unmanned aerial vehicles (UAVs) aligned with multistatic sensing in integrated sensing and communication (ISAC) systems can provide remarkable performance gains in sensing, by taking advantage of the cell-free massive multiple-input multiple-output (mMIMO) architecture. Under these considerations, in this paper, the achievable sensing signal-to-noise-plus-interference ratio (SINR) of a cell-free mMIMO ISAC UAV-based network is evaluated for two different deployments of UAVs, namely, mobile and tethered. In both scenarios, a transmit precoder that jointly optimizes the sensing and communication requirements subjected to power constraints is designed. Specifically, for the scenario with mobile UAVs, beyond the transmit precoding, we also optimize the position of the transmit UAVs through particle swarm optimization (PSO). The results show that, although tethered UAVs have a more efficient power allocation, the proposed position control algorithm for the mobile UAVs can achieve a superior gain in terms of sensing SINR.
Recent studies have shown that Large Language Models (LLMs) have the potential to process extremely long text. Many works only evaluate LLMs' long-text processing ability on the language modeling task, with perplexity (PPL) as the evaluation metric. However, in our study, we find that there is no correlation between PPL and LLMs' long-text understanding ability. Besides, PPL may only reflect the model's ability to model local information instead of catching long-range dependency. Therefore, only using PPL to prove the model could process long text is inappropriate. The local focus feature of PPL could also explain some existing phenomena, such as the great extrapolation ability of the position method ALiBi. When evaluating a model's ability in long text, we might pay more attention to PPL's limitation and avoid overly relying on it.
This position paper takes a broad look at Physics-Enhanced Machine Learning (PEML) -- also known as Scientific Machine Learning -- with particular focus to those PEML strategies developed to tackle dynamical systems' challenges. The need to go beyond Machine Learning (ML) strategies is driven by: (i) limited volume of informative data, (ii) avoiding accurate-but-wrong predictions; (iii) dealing with uncertainties; (iv) providing Explainable and Interpretable inferences. A general definition of PEML is provided by considering four physics and domain knowledge biases, and three broad groups of PEML approaches are discussed: physics-guided, physics-encoded and physics-informed. The advantages and challenges in developing PEML strategies for guiding high-consequence decision making in engineering applications involving complex dynamical systems, are presented.
The advancement of agricultural robotics holds immense promise for transforming fruit harvesting practices, particularly within the apple industry. The accurate detection and localization of fruits are pivotal for the successful implementation of robotic harvesting systems. In this paper, we propose a novel approach to apple detection and position estimation utilizing an object detection model, YOLOv5. Our primary objective is to develop a robust system capable of identifying apples in complex orchard environments and providing precise location information. To achieve this, we curated an autonomously labeled dataset comprising diverse apple tree images, which was utilized for both training and evaluation purposes. Through rigorous experimentation, we compared the performance of our YOLOv5-based system with other popular object detection models, including SSD. Our results demonstrate that the YOLOv5 model outperforms its counterparts, achieving an impressive apple detection accuracy of approximately 85%. We believe that our proposed system's accurate apple detection and position estimation capabilities represent a significant advancement in agricultural robotics, laying the groundwork for more efficient and sustainable fruit harvesting practices.
The field of web and mobile software frameworks is relatively mature, with a large variety of tools in different languages that facilitate traditional app development where data in a relational database is displayed and modified. Our position is that many current frameworks became popular during single server deployment of MVC architecture apps, and do not facilitate modern aspects of app development such as cloud computing and the incorporation of emerging technologies such as AI. We present a novel framework which accomplishes these purposes, Skeet, which was recently released to general use, alongside an initial evaluation. Skeet provides an app structure that reflects current trends in architecture, and tool suites that allow developers with minimal knowledge of AI internals to easily incorporate such technologies into their apps and deploy them.
This paper develops an underwater navigation solution that utilizes a strapdown inertial navigation system (SINS) and fuses a set of auxiliary sensors such as an acoustic positioning system, Doppler velocity log, depth meter, attitude meter, and magnetometer to accurately estimate an underwater vessel's position and orientation. The conventional integrated navigation system assumes Gaussian measurement noise, while in reality, the noises are non-Gaussian, particularly contaminated by heavy-tailed impulsive noises. To address this issue, and to fuse the system model with the acquired sensor measurements efficiently, we develop a square root polynomial chaos Kalman filter based on maximum correntropy criteria. The filter is initialized using acoustic beaconing to accurately locate the initial position of the vehicle. The computational complexity of the proposed filter is calculated in terms of flops count. The proposed method is compared with the existing maximum correntropy sigma point filters in terms of estimation accuracy and computational complexity. The simulation results demonstrate an improved accuracy compared to the conventional deterministic sample point filters.
Joint communication and sensing (JCS) is envisioned as an enabler of future 6G networks. One of the key features of these networks will be the use of extremely large aperture arrays (ELAAs) and high operating frequencies, which will result in significant near-field propagation effects. This unique property can be harnessed to improve sensing capabilities. In this paper, we focus on velocity sensing, as using ELAAs allows the estimation of not just the radial component but also the transverse component. We derive analytical performance bounds for both velocity components, demonstrating how they are affected by the different system parameters and geometries. These insights offer a foundational understanding of how near-field effects play in velocity sensing differently from the far field and from position estimate.
The complexity and dynamic nature of System of Systems (SoS) necessitate efficient communication mechanisms to ensure interoperability and collaborative functioning among constituent systems, termed holons. This paper proposes an innovative approach to enhance holon communication within SoS through the integration of Conversational Generative Intelligence (CGI) techniques. Our approach leverages advancements in CGI, specifically Large Language Models (LLMs), to enable holons to understand and act on natural language instructions. This fosters more intuitive human-holon interactions, improving social intelligence and ultimately leading to better coordination among diverse systems. This position paper outlines a conceptual framework for CGI-enhanced holon interaction, discusses the potential impact on SoS adaptability, usability and efficiency, and sets the stage for future exploration and prototype implementation.
Longstanding data labeling practices in machine learning involve collecting and aggregating labels from multiple annotators. But what should we do when annotators disagree? Though annotator disagreement has long been seen as a problem to minimize, new perspectivist approaches challenge this assumption by treating disagreement as a valuable source of information. In this position paper, we examine practices and assumptions surrounding the causes of disagreement--some challenged by perspectivist approaches, and some that remain to be addressed--as well as practical and normative challenges for work operating under these assumptions. We conclude with recommendations for the data labeling pipeline and avenues for future research engaging with subjectivity and disagreement.
Representation learning, and interpreting learned representations, are key areas of focus in machine learning and neuroscience. Both fields generally use representations as a means to understand or improve a system's computations. In this work, however, we explore surprising dissociations between representation and computation that may pose challenges for such efforts. We create datasets in which we attempt to match the computational role that different features play, while manipulating other properties of the features or the data. We train various deep learning architectures to compute these multiple abstract features about their inputs. We find that their learned feature representations are systematically biased towards representing some features more strongly than others, depending upon extraneous properties such as feature complexity, the order in which features are learned, and the distribution of features over the inputs. For example, features that are simpler to compute or learned first tend to be represented more strongly and densely than features that are more complex or learned later, even if all features are learned equally well. We also explore how these biases are affected by architectures, optimizers, and training regimes (e.g., in transformers, features decoded earlier in the output sequence also tend to be represented more strongly). Our results help to characterize the inductive biases of gradient-based representation learning. We then illustrate the downstream effects of these biases on various commonly-used methods for analyzing or intervening on representations. These results highlight a key challenge for interpretability $-$ or for comparing the representations of models and brains $-$ disentangling extraneous biases from the computationally important aspects of a system's internal representations.
Typical R-convolution graph kernels invoke the kernel functions that decompose graphs into non-isomorphic substructures and compare them. However, overlooking implicit similarities and topological position information between those substructures limits their performances. In this paper, we introduce Deep Hierarchical Graph Alignment Kernels (DHGAK) to resolve this problem. Specifically, the relational substructures are hierarchically aligned to cluster distributions in their deep embedding space. The substructures belonging to the same cluster are assigned the same feature map in the Reproducing Kernel Hilbert Space (RKHS), where graph feature maps are derived by kernel mean embedding. Theoretical analysis guarantees that DHGAK is positive semi-definite and has linear separability in the RKHS. Comparison with state-of-the-art graph kernels on various benchmark datasets demonstrates the effectiveness and efficiency of DHGAK. The code is available at Github (https://github.com/EWesternRa/DHGAK).
We propose Expected Work Search (EWS), a new game solving algorithm. EWS combines win rate estimation, as used in Monte Carlo Tree Search, with proof size estimation, as used in Proof Number Search. The search efficiency of EWS stems from minimizing a novel notion of Expected Work, which predicts the expected computation required to solve a position. EWS outperforms traditional solving algorithms on the games of Go and Hex. For Go, we present the first solution to the empty 5x5 board with the commonly used positional superko ruleset. For Hex, our algorithm solves the empty 8x8 board in under 4 minutes. Experiments show that EWS succeeds both with and without extensive domain-specific knowledge.
This position paper explores the benefits, drawbacks, and ethical considerations of incorporating synthetic personae in HCI research, particularly focusing on the customization challenges beyond the limitations of current Large Language Models (LLMs). These perspectives are derived from the initial results of a sub-study employing vignettes to showcase the existence of bias within black-box LLMs and explore methods for manipulating them. The study aims to establish a foundation for understanding the challenges associated with these models, emphasizing the necessity of thorough testing before utilizing them to create synthetic personae for HCI research.
This paper introduces UniGen, a novel approach to generating new traffic scenarios for evaluating and improving autonomous driving software through simulation. Our approach models all driving scenario elements in a unified model: the position of new agents, their initial state, and their future motion trajectories. By predicting the distributions of all these variables from a shared global scenario embedding, we ensure that the final generated scenario is fully conditioned on all available context in the existing scene. Our unified modeling approach, combined with autoregressive agent injection, conditions the placement and motion trajectory of every new agent on all existing agents and their trajectories, leading to realistic scenarios with low collision rates. Our experimental results show that UniGen outperforms prior state of the art on the Waymo Open Motion Dataset.
Sparse Vector Coding (SVC) has long been considered an encoding method that meets the URLLC QOS requirements. This encoding method has been widely studied and applied due to its low encoding and decoding complexity, no pilot transmission, resistance to inter-carrier interference, and low power consumption. However, due to the use of position indexing, the encoding essentially reduces the signal-to-noise ratio requirements by increasing the communication bandwidth, which also leads to low encoding efficiency and strong rigidity in decoding. Based on the sparse representation characteristics of SVC, we propose a joint sparse representation encoding, namely Sparse Matrix Coding (SMC). This encoding method utilizes multi-user information joint encoding, and the sparsity and sparse locations between users are shared.
We propose a hybrid formulation of the linear inverted pendulum model for bipedal locomotion, where the foot switches are triggered based on the center of mass position, removing the need for pre-defined footstep timings. Using a concept similar to reference spreading, we define nontrivial tracking error coordinates induced by our hybrid model. These coordinates enjoy desirable linear flow dynamics and rather elegant jump dynamics perturbed by a suitable extended class ${\mathcal K}_\infty$ function of the position error. We stabilize this hybrid error dynamics using a saturated feedback controller, selecting its gains by solving a convex optimization problem. We prove local asymptotic stability of the tracking error and provide a certified estimate of the basin of attraction, comparing it with a numerical estimate obtained from the integration of the closed-loop dynamics. Simulations on a full-body model of a real robot show the practical applicability of the proposed framework and its advantages with respect to a standard model predictive control formulation.
Multipath-based simultaneous localization and mapping (MP-SLAM) is a promising approach in wireless networks for obtaining position information of transmitters and receivers as well as information on the propagation environment. MP-SLAM models specular reflections of radio frequency (RF) signals at flat surfaces as virtual anchors (VAs), the mirror images of base stations (BSs). Conventional methods for MP-SLAM consider a single mobile terminal (MT) which has to be localized. The availability of additional MTs paves the way for utilizing additional information in the scenario. Specifically enabling MTs to exchange information allows for data fusion over different observations of VAs made by different MTs. Furthermore, cooperative localization becomes possible in addition to multipathbased localization. Utilizing this additional information enables more robust mapping and higher localization accuracy.
In the vast and dynamic landscape of urban settings, Traffic Safety Description and Analysis plays a pivotal role in applications ranging from insurance inspection to accident prevention. This paper introduces CityLLaVA, a novel fine-tuning framework for Visual Language Models (VLMs) designed for urban scenarios. CityLLaVA enhances model comprehension and prediction accuracy through (1) employing bounding boxes for optimal visual data preprocessing, including video best-view selection and visual prompt engineering during both training and testing phases; (2) constructing concise Question-Answer sequences and designing textual prompts to refine instruction comprehension; (3) implementing block expansion to fine-tune large VLMs efficiently; and (4) advancing prediction accuracy via a unique sequential questioning-based prediction augmentation. Demonstrating top-tier performance, our method achieved a benchmark score of 33.4308, securing the leading position on the leaderboard. The code can be found: https://github.com/alibaba/AICITY2024_Track2_AliOpenTrek_CityLLaVA