United States

Federated Learning (FL) is a distributed machine learning (ML) paradigm, in which multiple clients collaboratively train ML models without centralizing their local data. Similar to conventional ML pipelines, the client local optimization and server aggregation procedure in FL are sensitive to the hyperparameter (HP) selection. Despite extensive research on tuning HPs for centralized ML, these methods yield suboptimal results when employed in FL. This is mainly because their &quot;training-after-tuning&quot; framework is unsuitable for FL with limited client computation power. While some approaches have been proposed for HP-Tuning in FL, they are limited to the HPs for client local updates. In this work, we propose a novel HP-tuning algorithm, called Federated Population-based Hyperparameter Tuning (FedPop), to address this vital yet challenging problem. FedPop employs population-based evolutionary algorithms to optimize the HPs, which accommodates various HP types at both the client and server sides. Compared with prior tuning methods, FedPop employs an online  &quot;tuning-while-training&quot; framework, offering computational efficiency and enabling the exploration of a broader HP search space. Our empirical validation on the common FL benchmarks and complex real-world FL datasets, including full-sized Non-IID ImageNet-1K, demonstrates the effectiveness of the proposed method, which substantially outperforms the concurrent state-of-the-art HP-tuning methods in FL.

AAAI 2025

FedPop: Federated Population-based Hyperparameter Tuning

distributed machine learning

federated learning

Federated Learning (FL) is a distributed machine learning (ML) paradigm, in which multiple clients collaboratively train ML models without centralizing their local data. Similar to conventional ML pipelines, the client local optimization and server aggregation procedure in FL are sensitive to the hyperparameter (HP) selection. Despite extensive research on tuning HPs for centralized ML, these methods yield suboptimal results when employed in FL. This is mainly because their "training-after-tuning" framework is unsuitable for FL with limited client computation power. While some approaches have been proposed for HP-Tuning in FL, they are limited to the HPs for client local updates. In this work, we propose a novel HP-tuning algorithm, called Federated Population-based Hyperparameter Tuning (FedPop), to address this vital yet challenging problem. FedPop employs population-based evolutionary algorithms to optimize the HPs, which accommodates various HP types at both the client and server sides. Compared with prior tuning methods, FedPop employs an online  "tuning-while-training" framework, offering computational efficiency and enabling the exploration of a broader HP search space. Our empirical validation on the common FL benchmarks and complex real-world FL datasets, including full-sized Non-IID ImageNet-1K, demonstrates the effectiveness of the proposed method, which substantially outperforms the concurrent state-of-the-art HP-tuning methods in FL.

technical paper

We are pleased to announce the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI-25), which will be held in Philadelphia, Pennsylvania at the Pennsylvania Convention Center from February 25 to March 4, 2025.

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.

### [Invited Speakers](https://aaai.org/conference/aaai/aaai-25/aaai-25-invited-speakers/)

Register [here](https://aaai.org/conference/aaai/aaai-25/registration/)

The purpose of the AAAI conference series is to promote research in Artificial Intelligence (AI) and foster scientific exchange between researchers, practitioners, scientists, students, and engineers across the entirety of AI and its affiliated disciplines. AAAI-25 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and a range of other activities to be announced.



We study the fundamental problem of fairly dividing a set of indivisible items among agents with (general) monotone valuations. The notion of envy-freeness up to any item (EFX) is considered to be one of the most fascinating fairness concepts in this line of work. Unfortunately, despite significant efforts, existence of EFX allocations is a major open problem in fair division, thereby making the study of approximations and relaxations of EFX a natural line of research. Recently, Caragiannis et al. [2023] introduced a promising relaxation of EFX, called epistemic EFX (EEFX). We say an allocation to be EEFX if, for every agent, it is possible to shuffle the items in the remaining bundles so that she becomes “EFX-satisfied”. Caragiannis et al. [2023] prove existence and polynomial-time computability of EEFX allocations for additive valuations. A natural question asks what happens when we consider valuations more general than additive?

We address this important open question and answer it affirmatively by establishing the existence of EEFX allocations for an arbitrary number of agents with general monotone valuations. To the best of our knowledge, EEFX is the only known relaxation of EFX (beside EF1) to have such strong existential guarantees. Furthermore, we complement our existential result by proving computational and information-theoretic lower bounds. We prove that even for an arbitrary number of (more than one) agents with identical submodular valuations, it is PLS-hard to compute EEFX allocations and it requires exponentially-many value queries to do so.

Epistemic EFX Allocations Exist for Monotone Valuations

Federated prototype learning is in the spotlight as global prototypes are effective in enhancing the learning of local representation spaces, facilitating the ability to generalize the global model. However, when encountering domain-skewed data, conventional federated prototype learning is susceptible to two dilemmas: 1) Local prototypes obtained by averaging intra-class embedding carry domain-specific markers, the margins among aggregated global prototypes could be attenuated and detrimental to inter-class separation. 2) Local domain-skewed embedding may not exhibit a uniform distribution in Enclidean space, which is not conductive to the prototype-induced intra-class compactness. To address the two drawbacks, we go beyond conventional  paradigm of federated prototype learning, and propose learnable semantic anchors with hyperspherical contrast (FedLSA) for domain-skewed data. Specifically, we eschew the pattern of yielding prototypes via averaging intra-class embedding and directly learn a set of semantic anchors aided by the global semantic-aware classifier. Meanwhile, the margins between anchors are augmented via pulling apart varying anchors, ensuring decent inter-class separation. To guarantee that local domain-skewed representations can be uniformly distributed, local data is projected into the hyperspherical space, and the intra-class compactness is achieved by optimizing the contrastive loss derived from the von Mises-Fisher distribution. Finally, extensive experimental results on three multi-domain datasets show the superiority of the proposed FedLSA compared to state-of-the-art methods.

Beyond Federated Prototype Learning: Learnable Semantic Anchors with Hyperspherical Contrast for Domain-Skewed Data

The maximin optimisation problem, inspired by von Neumann’s work (von Neumann 1928) and widely applied in adversarial optimisation, has become a key research area in machine learning. Gradient Descent Ascent (GDA) is a common method for solving these problems but requires the payoff function to be differentiable, making it unsuitable for discrete or binary functions that often occur in game-theoretical scenarios. Co-evolutionary algorithms (CoEAs), which are derivative-free, offer an alternative to these problems. However, the theoretical understanding of CoEAs is still limited.

This paper provides the first rigorous runtime analysis of CoEAs with pairwise dominance on binary two-player zero-sum games (or maximin problems), specifically focusing on the DIAGONAL game. The mathematical analysis rigorously shows that the PDCoEA can efficiently find the optimum in polynomial runtime with high probability under low mutation rates and large population sizes. Empirical evidence also identifies an error threshold where higher mutation rates lead to inefficiency. In contrast, single-pair-individual algorithms, i.e.RLS-PD and (1+1)-CoEAs, fail to find the optimum in polynomial time. These findings highlight the importance of pairwise dominance, low mutation rates, and large populations in maintaining the “co-evolutionary arms race”.

Towards Runtime Analysis of Population-Based Co-evolutionary Algorithms on Sparse Binary Zero-Sum Game

Current neural networks often employ multi-domain-learning or attribute-injecting mechanisms to incorporate non-independent and identically distributed (non-IID) information for text understanding tasks by capturing individual characteristics and the relationships among samples. However, the extent of the impact of non-IID information and how these methods affect pre-trained language models (PLMs) remains unclear. This study revisits the assumption that non-IID information enhances PLMs to achieve performance improvements from a Bayesian perspective, which unearths and integrates non-IID and IID features. Furthermore, we proposed a multi-attribute multi-grained framework for PLM adaptations (M2A), which combines multi-attribute and multi-grained views to mitigate uncertainty in a lightweight manner. We evaluate M2A through prevalent text-understanding datasets and demonstrate its superior performance, mainly when data are implicitly non-IID, and PLMs scale larger. Our code is publicly available at https://anonymous.4open.science/r/M2A-62EA/.

Multi-attribute Multi-grained Adaptation of Pre-trained Language Models for Text Understanding from Bayesian Perspective

Multi-Robot Coverage problems have been extensively studied in robotics, planning and multi-agent systems. In this work, we consider the coverage problem when there are constraints on the proximity (e.g., maximum distance between the agents, or a blue agent must be adjacent to a red agent) and the movement (e.g., terrain traversability and material load capacity) of the robots.
Such constraints naturally arise in many real-world applications, e.g. in search-and-rescue and maintenance operations. Given such a setting, the goal is to compute a covering tour of the graph with a minimum number of steps, and that adheres to the proximity and movement constraints.   
For this problem, our contributions are four: (i) a formal formulation of the problem, (ii) an exact algorithm that is FPT in parameters $\|\mathcal{F}\|$, $d$ and $\textsf{tw}$ - the set of robot formations that encode the proximity constraints, the maximum nodes degree, and the tree-width of the graph, respectively, (iii) for the case that the graph is a tree: a PTAS approximation scheme, that given an $\varepsilon$ produces a tour that is within a $1+\varepsilon\cdot \textsf{error}(\|\mathcal{F}\|,d))$ of the optimal one, and the computation runs in time $\textsf{poly}(n) \cdot h(\frac{1}{\varepsilon},\mathcal{F})$.
(iv) for the case that the graph is a tree, with $k=3$ robots, and the constraint is that all agents are connected: a PTAS scheme with multiplicative approximation error of $1+\mathcal{O}(\varepsilon)$, independent of $d$.

Heterogeneous Multi-Robot Graph Coverage with Proximity and Movement Constraints

Uncertainty estimation has been widely applied for trustworthy automatic speech recognition (ASR) systems across a training stage and an inference stage. At the training stage, previous studies show that uncertainty can facilitate self-training by filtering out unlabeled data samples with high uncertainty. However, the current sequence-level uncertainty estimation method for connectionist temporal classification (CTC) based ASR models drops the output probability information and depends only on the textual distance of decoded predictions. In this study, we argue that this results in limited performance improvement and propose a novel output probability-based sequence-level uncertainty estimation method. We also categorize uncertainty as pseudo-label uncertainty and in-training uncertainty for the self-training process. Finally, we present uncertainty-aware self-training for CTC-based ASR models and experimentally show the effectiveness of the proposed method compared to the baselines.

Uncertainty-Aware Self-Training for CTC-Based Automatic Speech Recognition

Recent works on remote PhotoPlethysmoGraphy (rPPG) estimation typically use techniques like CNNs and Transformers to encode implicit features from facial videos for prediction.
These methods learn to directly map facial videos to the static values of rPPG signals, overlooking the inherent dynamic characteristics of rPPG sequence. Moreover, the rPPG signal is extremely weak and highly susceptible to interference from various sources of noise, including illumination conditions, head movements, and variations in skin tone. To address these limitations, we propose a p、Physiology-based dynamicity disentangled Diffusion (PhysDiff) model particularly designed for robust rPPG estimation. PhysDiff leverages the diffusion model to learn the distribution of quasi-periodic rPPG signal and uses a \textit{dynamicity disentanglement strategy} to capture two dynamic characteristics in temporal rPPG signal, \ie, trend and amplitude. This disentanglement is motivated by the underlying dynamic physiological processes of vasodilation and vasoconstriction, ensuring a more precise representation of the rPPG signal. The disentangled components are then used as pivotal conditions in the proposed \textit{spatial-temporal hybrid denoiser} for rPPG reconstruction. Besides, we introduce a \textit{periodicity-based multi-hypothesis selection strategy} in model inference, which compares the natural periodicity of multiple generated rPPG hypotheses with the ground truth and selects the most favorable one as the final prediction. Extensive experiments on four datasets demonstrate that our PhysDiff significantly outperforms prior methods on both intra-dataset and cross-dataset testing. The code will be released after the double-blind review.

PhysDiff: Physiology-based Dynamicity Disentangled Diffusion Model for Remote Physiological Measurement

Predicting Click-Through Rates is a crucial function within recommendation and advertising platforms, as the output of CTR prediction determines the order of items shown to users. The Embedding \& MLP paradigm has become a standard approach for industrial recommendation systems and has been widely deployed. However, this paradigm suffers from cold-start problems, where there is either no or only limited user action data available, leading to poorly learned ID embeddings. The cold-start problem hampers the performance of new items. To address this problem, we designed a novel diffusion model to generate a warmed-up embedding for new items.
Specifically, we define a novel diffusion process between the ID embedding space and the side information space.
In addition, we can derive a sub-sequence from the diffusion steps to expedite training, given that our diffusion model is non-Markovian.
Our diffusion model is supervised by both the variational inference and binary cross-entropy objectives, enabling it to generate warmed-up embeddings for items in both the cold-start and warm-up phases. Additionally, we have conducted extensive experiments on three recommendation datasets. The results confirmed the effectiveness of our approach.

Addressing Cold-Start Problem in Click-Through Rate Prediction via Supervised Diffusion Modeling

Hard combinatorial optimization problems, often mapped to Ising models, promise potential solutions with quantum advantage but are constrained by limited qubit counts in near-term devices. We present an innovative quantum-inspired framework that dynamically compresses large Ising models to fit available quantum hardware of different sizes. Thus, we aim to bridge the gap between large-scale optimization and current hardware capabilities. Our method leverages a physics-inspired GNN architecture to capture complex interactions in Ising models and accurately predict alignments among neighboring spins (aka qubits) at ground states. By progressively merging such aligned spins, we can reduce the model size while preserving the underlying optimization structure. It also provides a natural trade-off between the solution quality and size reduction, meeting different hardware constraints of quantum computing devices. Extensive numerical studies on Ising instances of diverse topologies show that our method can reduce instance size at multiple levels with virtually no losses in solution quality on the latest D-wave quantum annealers.

Scalable Quantum-Inspired Optimization Through Dynamic Qubit Compression

Controversial contents largely inundate the Internet, infringing various cultural norms and child protection standards. Traditional Image Content Moderation (ICM) models fall short in producing precise moderation decisions for diverse standards, while recent multimodal large language models (MLLMs), when adopted to general rule-based ICM, often produce classification and explanation results that are inconsistent with human moderators. Aiming at flexible, explainable and accurate ICM, we design a novel rule-based dataset generation pipeline, decomposing concise human-defined rules and leveraging well-designed multi-stage prompts to enrich short explicit image annotations. Our ICM-Instruct dataset includes detailed moderation classification, explanation, and question-answer pairs. Building on it, we create our ICM-Assistant model in the framework of rule-based ICM, making it readily applicable in real practice. Our ICM-Assistant model demonstrates exceptional performance and flexibility.  Specifically, it significantly outperforms existing approaches on various moderation tasks, rules, and data sources (e.g., user-generated, AI-generated, etc.), improving both the moderation classification (33.7% on average) and moderation explanation quality (26.6% on average) consistently over the existing MLLMs.

FedPop: Federated Population-based Hyperparameter Tuning

Downloads

Next from AAAI 2025

Epistemic EFX Allocations Exist for Monotone Valuations

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES