Robot soccer, in its full complexity, poses an unsolved research challenge. Current solutions heavily rely on engineered heuristic strategies, which lack robustness and adaptability. Deep reinforcement learning has gained significant traction in various complex robotics tasks such as locomotion, manipulation, and competitive games (e.g., AlphaZero, OpenAI Five), making it a promising solution to the robot soccer problem. This paper introduces MARLadona. A decentralized multi-agent reinforcement learning (MARL) training pipeline capable of producing agents with sophisticated team play behavior, bridging the shortcomings of heuristic methods. Furthermore, we created an open-source multi-agent soccer environment. Utilizing our MARL framework and a modified global entity encoder (GEE) as our core architecture, our approach achieves a 66.8 % win rate against HELIOS agent, which employs a state-of-the-art heuristic strategy. In addition, we provided an in-depth analysis of the policy behavior and interpreted the agent's intention using the critic network.

Serving machine learning models on edge, mobile, and embedded devices places stringent requirements on inference latency. From operating a real enterprise service, we observed that even a fully optimized model could lead to severe violations of latency objectives when the load surges. A straightforward and mature approach is to auto-scale multiple models to balance the load. However, unlike cloud clusters, edge or mobile devices usually cannot afford to deploy multiple model replicas. Therefore, in this paper, we explore a new idea: in addition to the original model, we deploy one (or more) heterogeneous model(s) with much smaller resource overhead on the device, and perform load balancing among all models. We overcame the technical challenges posed by performance dynamics and developed InferRouter based on queuing theory. We implement and evaluate InferRouter on three real on-device inference systems, covering mobile sensing, video analytics, and natural language processing applications. Experimental results show that compared with strong baselines, InferRouter can decrease 85.2% P99 latency (5.8x faster) and improve 5.9% accuracy on the mobile workload. For a traffic video analytics task, InferRouter achieves 55.1% higher accuracy with zero deadline misses. InferRouter also shows its advantages in saving resources compared with auto-scaling and offloading approaches.

Spatio-temporal point processes (STPPs) are potent mathematical tools for modeling and predicting events with both temporal and spatial features. Despite their versatility, most existing methods for learning STPPs either assume a restricted form of the spatio-temporal distribution, or suffer from inaccurate approximations of the intractable integral in the likelihood training objective. These issues typically arise from the normalization term of the probability density function. Moreover, existing works only provide point prediction for events without quantifying their uncertainty, such as confidence intervals for the event’s arrival time and confidence regions for the event’s location, which is crucial given the considerable randomness of the data. To tackle these challenges, we introduce SMASH: a Score MAtching-based pSeudolikeliHood estimator for learning marked STPPs. Specifically, our framework adopts a normalization-free objective by estimating the pseudolikelihood of marked STPPs through score-matching and predicts confidence intervals/regions for event time and location by generating samples through a score-based sampling algorithm. The superior performance of our proposed framework is demonstrated through extensive experiments on both point and confidence interval/region prediction of events.

Reinforcement learning from human feedback (RLHF) provides a principled framework for aligning AI systems with human preference data. For various reasons, e.g., personal bias, context ambiguity, lack of training, etc, human annotators may give incorrect or inconsistent preference labels. To tackle this challenge, we propose a robust RLHF approach -- $R^3M$, which models the potentially corrupted preference label as sparse outliers. Accordingly, we formulate the robust reward learning as an $\ell_1$-regularized maximum likelihood estimation problem. Computationally, we develop an efficient alternating optimization algorithm, which only incurs negligible computational overhead compared with the standard RLHF approach. Theoretically, we prove that under proper regularity conditions, $R^3M$ can consistently learn the underlying reward and identify outliers, provided that the number of outlier labels scales sublinearly with the preference sample size. Furthermore, we remark that $R^3M$ is versatile and can be extended to various preference optimization methods, including direct preference optimization (DPO). Our experiments on robotic control and natural language generation with large language models (LLMs) show that $R^3M$ improves robustness of the reward against several types of perturbations to the preference data.

Reinforcement learning from human feedback (RLHF) is a prevalent approach to align AI systems with human values by learning rewards from human preference data. Due to various reasons, however, such data typically takes the form of rankings over pairs of trajectory segments, which fails to capture the varying strengths of preferences across different pairs. In this paper, we propose a novel adaptive preference loss, underpinned by distributionally robust optimization (DRO), designed to address this uncertainty in preference strength. By incorporating an adaptive scaling parameter into the loss for each pair, our method increases the flexibility of the reward function. Specifically, it assigns small scaling parameters to pairs with ambiguous preferences, leading to more comparable rewards, and large scaling parameters to those with clear preferences for more distinct rewards. Computationally, our proposed loss function is strictly convex and univariate with respect to each scaling parameter, enabling its efficient optimization through a simple second-order algorithm. Our method is versatile and can be readily adapted to various preference optimization frameworks, including direct preference optimization (DPO). Our experiments with robotic control and natural language generation with large language models (LLMs) show that our method not only improves policy performance but also aligns reward function selection more closely with policy optimization, simplifying the hyperparameter tuning process.

Deep ensemble learning has been widely adopted to boost accuracy through combing outputs from multiple deep models prepared for the same task. However, the extra computation and memory cost it entails could impose an unacceptably high deadline miss rate in latency-sensitive tasks. Conventional approaches, including ensemble selection, focus on accuracy while ignoring deadline constraints, and thus cannot smartly cope with bursty query traffic and queries with different hardness. This paper explores redundancy in deep ensemble model inference and presents Schemble, a query difficulty-dependent task scheduling framework. Schemble treats ensemble inference progress as multiple base model inference tasks and schedules tasks for queries based on their difficulty and queuing status. We evaluate Schemble on real-world datasets, considering intelligent Q&A system, video analysis and image retrieval as the running applications. Experimental results show that Schemble achieves a 5× lower deadline miss rate and improves the accuracy by 30.8% given deadline constraints.

Transformer Hawkes process models have shown to be successful in modeling event sequence data. However, most of the existing training methods rely on maximizing the likelihood of event sequences, which involves calculating some intractable integral. Moreover, the existing methods fail to provide uncertainty quantification for model predictions, e.g., confidence interval for the predicted event’s arrival time. To address these issues, we propose SMURF-THP, a score-based method for learning Transformer Hawkes process and quantifying prediction uncertainty. Specifically, SMURF-THP learns the score function of the event’s arrival time based on a score-matching objective that avoids the intractable computation. With such a learnt score function, we can sample arrival time of events from the predictive distribution. This naturally allows for the quantification of uncertainty by computing confidence intervals over the generated samples. We conduct extensive experiments in both event type prediction and uncertainty quantification on time of arrival. In all the experiments, SMURF-THP outperforms existing likelihood-based methods in confidence calibration while exhibiting comparable prediction accuracy.

The success of many web services relies on the large-scale domain-specific high-quality labeled dataset. Insufficient public datasets motivate us to reduce the cost of data labeling while maintaining high accuracy in support of intelligent web applications. The rule-based method and the learning-based method are common techniques for labeling. In this work, we study how to utilize the rule-based and learning-based methods for resource-effective text labeling. We propose CoTel, the first ontology-neural co-enhanced framework for text labeling. We propose critical ontology extraction in the rule-based module and ontology-enhanced loss prediction in the learning-based module. CoTel can integrate explicit labeling rules and implicit labeling models and make them help each other to improve resource efficiency in text labeling tasks. We evaluate CoTel on both public datasets and real applications with three different tasks. Compared with the baseline, CoTel can reduce the time cost by 64.75% (a 2.84× speedup) and the number of labeling by 62.07%.

In many real-world problems, first-order (FO) derivative evaluations are too expensive or even inaccessible. For solving these problems, zeroth-order (ZO) methods that only need function evaluations are often more efficient than FO methods or sometimes the only options. In this paper, we propose a novel zeroth-order inexact augmented Lagrangian method (ZO-iALM) to solve black-box optimization problems, which involve a composite (i.e., smooth+nonsmooth) objective and functional constraints. This appears to be the first work that develops an iALM-based ZO method for functional constrained optimization and meanwhile achieves query complexity results matching the best-known FO complexity results up to a factor of variable dimension. With an extensive experimental study, we show the effectiveness of our method. The applications of our method span from classical optimization problems to practical machine learning examples such as resource allocation in sensor networks and adversarial example generation.

First-order methods have been studied for nonlinear constrained optimization within the framework of the augmented Lagrangian method (ALM) or penalty method. We propose an improved inexact ALM (iALM) and conduct a unified analysis for nonconvex problems with either convex or nonconvex constraints. Under certain regularity conditions (that are also assumed by existing works), we show an $\tilde{O}(\varepsilon^{-\frac{5}{2}})$ complexity result for a problem with a nonconvex objective and convex constraints and an $\tilde{O}(\varepsilon^{-3})$ complexity result for a problem with a nonconvex objective and nonconvex constraints, where the complexity is measured by the number of first-order oracles to yield an $\varepsilon$-KKT solution. Both results are the best known. The same-order complexity results have been achieved by penalty methods. However, two different analysis techniques are used to obtain the results, and more importantly, the penalty methods generally perform significantly worse than iALM in practice. Our improved iALM and analysis close the gap between theory and practice. Numerical experiments on nonconvex problems with convex or nonconvex constraints are provided to demonstrate the effectiveness of our proposed method.

Modern data acquisition routinely produce massive amounts of event sequence data in various domains, such as social media, healthcare, and financial markets. These data often exhibit complicated short-term and long-term temporal dependencies. However, most of the existing recurrent neural network based point process models fail to capture such dependencies, and yield unreliable prediction performance. To address this issue, we propose a Transformer Hawkes Process (THP) model, which leverages the self-attention mechanism to capture long-term dependencies and meanwhile enjoys computational efficiency. Numerical experiments on various datasets show that THP outperforms existing models in terms of both likelihood and event prediction accuracy by a notable margin. Moreover, THP is quite general and can incorporate additional structural knowledge. We provide a concrete example, where THP achieves improved prediction performance for learning multiple point processes when incorporating their relational information.