Singapore

Knowledge distillation (KD) is a key technique for compressing large-scale language models (LLMs), yet prevailing logit-based methods typically employ static strategies that are misaligned with the dynamic learning process of student models. These methods typically treat all tokens indiscriminately and apply a single, fixed temperature, resulting in suboptimal knowledge transfer. To address these limitations, we propose LLM-Oriented Token-Adaptive Knowledge Distillation (AdaKD), a novel framework that adapts the distillation process to the real-time learning state of each token. AdaKD consists of two synergistic modules driven by a unified token difficulty metric. First, our Loss-Driven Adaptive Token Focusing (LATF) module dynamically adjusts the distillation focus by monitoring the student&#39;s learning stability, concentrating computational resources on the most valuable tokens at each training phase. Second, we introduce Inverse Difficulty Temperature Scaling (IDTS), a counterintuitive yet effective token-level temperature strategy. It employs low temperatures for difficult tokens for targeted error correction, and high temperatures for easy tokens to encourage students to learn from the teacher&#39;s complete and smooth output distribution, thereby enhancing generalization. As a plug-and-play framework, AdaKD can consistently improve the performance of various distillation methods on multiple model architectures and benchmarks.

AAAI 2026

LLM-Oriented Token-Adaptive Knowledge Distillation

nlp: learning & optimization for nlp

ml: efficient ml / green ai

nlp: (large) language models

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-23 is the Thirty-Seventh AAAI Conference on Artificial Intelligence. The theme of this conference is to create collaborative bridges within and beyond AI. Like previous AAAI conferences, AAAI-23 will feature technical paper presentations, special tracks, invited speakers, workshops, tutorials, poster sessions, senior member presentations, competitions, and exhibit programs, and two new activities: a Bridge Program and a Lab Program. Many of these activities are tailored to the theme of bridges and all are selected according to the highest standards, with additional programs for students and young researchers. 
AAAI is providing you with a conference planner, which you can use to help organize your itinerary of activities. This includes talks to attend in person, talks to attend remotely, breaks with colleagues and your site seeing activities. To access this conference planner, please go to [https://aaai-2023.takemobi.io](https://aaai-2023.takemobi.io).

In order to access this site, you need to register. If you haven't already, please register [here](https://aaai.org/Conferences/AAAI-23/registration/).


AAAI 2023

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.

Knowledge distillation (KD) is a key technique for compressing large-scale language models (LLMs), yet prevailing logit-based methods typically employ static strategies that are misaligned with the dynamic learning process of student models. These methods typically treat all tokens indiscriminately and apply a single, fixed temperature, resulting in suboptimal knowledge transfer. To address these limitations, we propose LLM-Oriented Token-Adaptive Knowledge Distillation (AdaKD), a novel framework that adapts the distillation process to the real-time learning state of each token. AdaKD consists of two synergistic modules driven by a unified token difficulty metric. First, our Loss-Driven Adaptive Token Focusing (LATF) module dynamically adjusts the distillation focus by monitoring the student's learning stability, concentrating computational resources on the most valuable tokens at each training phase. Second, we introduce Inverse Difficulty Temperature Scaling (IDTS), a counterintuitive yet effective token-level temperature strategy. It employs low temperatures for difficult tokens for targeted error correction, and high temperatures for easy tokens to encourage students to learn from the teacher's complete and smooth output distribution, thereby enhancing generalization. As a plug-and-play framework, AdaKD can consistently improve the performance of various distillation methods on multiple model architectures and benchmarks.

technical paper

We are pleased to announce the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26), which will be held in Singapore EXPO from January 20 to January 27, 2026.

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-26 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.

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to AAAI 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://aaai.org/conference/aaai/aaai-26/registration/) first.

**Online Registration Form**: https://aaai.getregistered.net/conference-2026 

Registration Required

We are pleased to announce the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26), which will be held in Singapore EXPO from January 20 to January 27, 2026.

Dynamic Vision Sensor (DVS) asynchronously records sparse events triggered by changes in pixel intensity, offering high temporal resolution and low latency. Existing frame-based methods process event data densely, violating its inherent sparsity and introducing computational redundancy. While asynchronous models preserve the event stream's native format, they often neglect spatial information, compromising their adaptability and efficiency. To address these limitations, we propose a Spatiotemporally Separated Sparse Network (S3Net) for efficient event stream encoding and learning. Specifically, we employ a learnable sparse encoding scheme to construct a voxel-structured representation that effectively extracts spatiotemporal relationships among event data. After that, we propose a dual-branch architecture to capture localized spatial dependencies and dynamic temporal patterns of event data. By explicitly decoupling spatial and temporal modeling, S3Net enables end-to-end asynchronous processing of variable-length event sequences, achieving both strong representational capacity and high computational efficiency. Experimental results on six event-based datasets demonstrate that S3Net achieves state-of-the-art performance. Compared to frame-based methods, it significantly reduces computational overhead and model complexity, while also outperforming existing asynchronous approaches in inference speed without compromising accuracy. Extensive experiments across six event-based datasets show that S3Net establishes new state-of-the-art performance. Our method reduces computational costs by 35% and model parameters by 27% compared to frame-based approaches, while delivering 1.58× faster inference than existing point-based methods at comparable accuracy levels.

S3Net: Spatiotemporally Separated Sparse Network for Neuromorphic Vision Processing

Incomplete multi-modal emotion recognition (IMER) aims at understanding human intentions and sentiments by comprehensively exploring the partially observed multi-source data. Although the multi-modal data is expected to provide more abundant information, the performance gap and modality under-optimization problem hinder effective multi-modal learning in practice, and are exacerbated in the confrontation of the missing data. To address this issue, we devise a novel Cross-modal Prompting (ComP) method, which emphasizes coherent information by enhancing modality-specific features and improves the overall recognition accuracy by boosting each modality's performance. Specifically, a progressive prompt generation module with a dynamic gradient modulator is proposed to produce concise and consistent modality semantic cues. Meanwhile, cross-modal knowledge propagation selectively amplifies the consistent information in modality features with the delivered prompts to enhance the discrimination of the modality-specific output. Additionally, a coordinator is designed to dynamically re-weight the modality outputs as a complement to the balance strategy to improve the model's efficacy. Extensive experiments on 4 datasets with 7 SOTA methods under different missing rates validate the effectiveness of our proposed method.

Cross-modal Prompting for Balanced Incomplete Multi-modal Emotion Recognition

Since a building's floorplan remains consistent over time and is inherently robust to changes in visual appearance, visual Floorplan Localization (FLoc) has received increasing attention from researchers. However, as a compact and minimalist representation of the building's layout, floorplans contain many repetitive structures (e.g., hallways and corners), thus easily result in ambiguous localization. Existing methods either pin their hopes on matching 2D structural cues in floorplans or rely on 3D geometry-constrained visual pre-trainings, ignoring the richer contextual information provided by visual images. In this paper, we suggest using broader visual scene context to empower FLoc algorithms with scene layout priors to eliminate localization uncertainty. In particular, we propose an unsupervised learning technique with clustering constraints to pre-train a room discriminator on self-collected unlabeled room images. Such a discriminator can empirically extract the hidden room type of the observed image and distinguish it from other room types. By injecting the scene context information summarized by the discriminator into an FLoc algorithm, the room style knowledge is effectively exploited to guide definite visual FLoc. We conducted sufficient comparative studies on two standard visual Floc benchmarks. Our experiments show that our approach outperforms state-of-the-art methods and achieves significant improvements in robustness and accuracy.

Perspective from a Broader Context: Can Room Style Knowledge Help Visual Floorplan Localization?

Knowledge Tracing (KT) aims to mine students’ evolving knowledge states and predict their future question-answering performance. Existing methods based on heterogeneous information networks (HINs) are prone to introducing noises due to manual or random selection of meta-paths and lack necessary quality assessment of meta-path instances. Conversely, recent large language models (LLMs)-based methods ignore the rich information across students, and both paradigms struggle to deliver consistently accurate and evidence-based explanations. To address these issues, we propose an innovative framework, HIN-LLM Synergistic Enhanced Knowledge Tracing (HISE-KT), which seamlessly integrates HINs with LLMs. HISE-KT first builds a multi-relationship HIN containing diverse node types to capture the structural relations through multiple meta-paths. The LLM is then employed to intelligently score and filter meta-path instances and retain high-quality paths, pioneering automated meta-path quality assessment. Inspired by educational psychology principles, a similar student retrieval mechanism based on meta-paths is designed to provide a more valuable context for prediction. Finally, HISE-KT uses a structured prompt to integrate the target student's history with the retrieved similar trajectories, enabling the LLM to generate not only accurate predictions but also evidence-backed, explainable analysis reports. Experiments on four public datasets show that HISE-KT outperforms existing KT baselines in both prediction performance and interpretability.

HISE-KT: Synergizing Heterogeneous Information Networks and LLMs for Explainable Knowledge Tracing with Meta-Path Optimization

To identify the root causes of attacks, behavior abstraction (BA) converts audit logs into multiple behavior graphs and finds similar ones, which has proven effective in bridging the semantic gap and reducing manual workload. Existing works fail to achieve both interpretability and generalization, while also exhibiting limited robustness when facing adversarial attacks. In this paper, we give the first attempt at interpretable and robust behavior abstraction and propose a novel method called 
$\textit{\textbf{E}nvironment-\textbf{D}isentangled \textbf{H}eterogeneous \textbf{G}raph \textbf{N}eural \textbf{N}etwork (\textbf{EDHGNN})}$. Motivated by Information Bottleneck (IB) principle, we propose a Heterogeneous Subgraph Disentanglement (HSD) module to disentangle label-relevant and environmental subgraphs through single optimization. We also introduce an Adapted Graph-Level Attention (AGLA) module to extract minimal sufficient representations from label-relevant subgraphs, a Label-Guided Graph Reconstructor (LGGR) to maximize environmental information coverage via reconstruction, and a Relevance Discriminator (RD) to enhance disentanglement quality. Additionally, we construct a new dataset contains ground-truth explanations and 4,160 behavior graphs. Extensive experiments demonstrate that EDHGNN outperforms the state-of-the-art methods in terms of interpretability and robustness against
adversarial attacks.

Interpretable and Robust Behavior Abstraction via Environment-Disentangled Heterogeneous Graph

The Multi-Agent Path Finding (MAPF) problem is a computationally challenging task that involves coordinating collision-free trajectories for multiple cooperative agents. Although existing methods address corridor symmetry, where agents encounter repeated bidirectional conflicts in constrained environments, they typically focus exclusively on pairwise agent interactions. Our observations reveal that such pairwise symmetry frequently arises when multiple agents traverse shared corridors, necessitating repeated applications of the corridor reasoning technology over extended durations. To overcome this limitation, we propose a multi-agent corridor reasoning (MAC) technology capable of resolving group-level corridor symmetry in a single optimization step. Our theoretical analysis demonstrates that this technology preserves the completeness and optimality guarantees of Conflict-Based Search (CBS). By integrating MAC technology with CBSH-RTC, we developed CBSH-MACRT, which significantly outperforms state-of-the-art algorithms (CBSH-RTC and CBSH with mutex propagation) on standardized MAPF benchmarks, improving success rates by 8–40\% and cutting runtimes by 14–67\%.

Multi-Agent Corridor Reasoning for Multi-Agent Path Finding

This paper proposes a framework for improving the operational efficiency of automated storage systems under uncertainty.
Recent years have seen a rise in automated grid-based storage for uniform-sized \emph{loads}, e.g., containers, pallets, totes.
Such systems face a fundamental tradeoff between maximizing space utilization and minimizing costly load relocation efforts during storage and retrieval operations. 
The focus here is on a setting with unique loads that can move along cardinal directions using a single mobile manipulator, such as a robot.
The setting consists of two distinct phases, common in some logistics applications, such as last-mile distribution centers and shipyards: i) storage of all the loads, followed by ii) their retrieval.
The goal is to minimize relocations for both phases, especially when the storage system is at capacity.
Previous efforts have shown that with known storage and retrieval orders, zero relocations can be achieved for storage at full capacity, provided that the size of the opening through which loads are stored and retrieved (grid width) is at least 3.

In realistic scenarios, however, schedules may be uncertain, i.e., loads may be stored or retrieved out of order, rendering previous approaches suboptimal.
The model of uncertainty in this work assumes that any two departing loads may depart out of order if they are originally at most $k$ positions apart. Under this model, this work
generalizes the previous result and proves that a grid width of $\Theta(k)$ is necessary and sufficient for eliminating relocations via robust storage arrangements.
An efficient solver is introduced to find such robust arrangements.
Furthermore, when relocations become inevitable, such as when loads are retrieved out of order by more than $k$, a strategy is introduced that effectively minimizes total relocations.
Extensive experiments show that, for $k$ up to half the grid width, the proposed storage and retrieval approaches essentially eliminate relocations.
For high uncertainty, i.e., $k$ values up to the full grid width, relocations are reduced by $50\%+$.

Robust Out-of-Order Retrieval for Grid-Based Storage at Maximum Capacity

Chain-of-Thought (CoT) reasoning is a critical capability for large language models (LLMs), enabling them to tackle complex multi-step tasks. While base LLMs, pre-trained on general text corpora, often struggle with reasoning due to a lack of specialized training, recent studies reveal their latent reasoning potential tied to hidden states. However, existing hidden state manipulation methods, such as linear activation steering, suffer from limitations due to their rigid and unconstrained nature, often leading to distribution shifts and degraded text quality. In this work, we propose a novel approach for eliciting CoT reasoning from base LLMs through hidden state manipulation grounded in probabilistic conditional generation. By reformulating the challenge as an optimization problem with a balanced likelihood and prior regularization framework, our method guides hidden states toward reasoning-oriented trajectories while preserving linguistic coherence. Extensive evaluations across mathematical, commonsense, and logical reasoning benchmarks demonstrate that our approach consistently outperforms existing steering methods, offering a theoretically principled and effective solution for enhancing reasoning capabilities in base LLMs.

Eliciting Chain-of-Thought in Base LLMs via Gradient-Based Representation Optimization

Large Language Models (LLMs) have recently emerged as a leading approach for multivariate time series forecasting. However, their effectiveness is hampered by a fundamental architectural mismatch: the permutation-invariant self-attention of Transformers lacks inductive biases for the strict temporal order and complex cross-variable dependencies inherent in time series. Existing methods often sidestep this issue with input-level alignment techniques rather than endowing the model itself with structural awareness. To address this gap, we introduce \textbf{GraFT} (\textbf{Gra}ph-infused \textbf{F}orecasting \textbf{T}ransformer), a framework that systematically embeds relational priors into a pre-trained backbone by constructing a heterogeneous patch relation graph, which represents both universal temporal principles with static edges and instance-specific patterns with dynamic adaptive edges. To process this multi-relational structure, a relational graph convolutional network generates structure-aware representations, which are infused into the patch embeddings to provide explicit structural guidance to the Transformer's attention mechanism. Extensive experiments show that GraFT achieves state-of-the-art performance on long-term forecasting and zero-shot learning, outperforming leading LLM-based methods on eight standard benchmarks with an average Mean Squared Error (MSE) reduction of 14.4\%.

GraFT: Infusing Pre-trained Transformers with Relational Structure for Time Series Forecasting

In the referenced manuscript, we introduced SUNT, to our
knowledge the richest open urban mobility dataset,
integrating continuously collected passenger and vehicle
data across three public transport systems in Salvador,
Brazil, namely regular buses, subway, and BRT. SUNT covers
daily activity for about 700,000 passengers and
approximately 2,000 vehicles operating over nearly 400
lines and close to 3,000 stops and stations, with
collection from March 2024 to March 2025 at subminute
resolution. The article details data acquisition and
organization, including technical validation through
descriptive statistics and quality checks, as well as the
implementation of a trip chaining to build
Origin–Destination flows. We present a large scale, real
world testbed for reasoning over graphs, sequences, and
uncertainty in continuous time, supporting research in
scalable online learning, distribution shift adaptation,
and decision making under constraints across logistics,
energy, and other infrastructure domains. At the conference
we will show results with new methods and recent state of
the art in Graph Neural Networks, Transformer based and
Recurrent models, Reinforcement Learning, Multiagent
methods, Continual Learning, and Foundation models.
Researchers across AI can use SUNT and the accompanying
models as a shared benchmark, enabling comparable,
reproducible, and transferable evaluation and accelerating
progress in temporal modeling, relational reasoning, and
decision making under uncertainty.

Downloads

Next from AAAI 2026

S3Net: Spatiotemporally Separated Sparse Network for Neuromorphic Vision Processing

Similar lecture

Curriculum Temperature for Knowledge Distillation

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from AAAI 2026

S3Net: Spatiotemporally Separated Sparse Network for Neuromorphic Vision Processing

Similar lecture

Curriculum Temperature for Knowledge Distillation

Downloads