28 accepted papers
Research on continual learning for LLMs seeks to acquire new skills without catastrophic forgetting of established prior knowledge. However, domain-specific fine-tuning still triggers severe, long-tailed forgetting issues even under narrow updates, particularly when the pre-training data is inaccess…
KV caches, typically used only to speed up autoregressive decoding, encode contextual information that can be reused for downstream tasks at no extra cost. We propose treating the KV cache as a lightweight representation, eliminating the need to recompute or store full hidden states. Despite being w…
Formal verification provides strong guarantees of software correctness, but its adoption is limited by the high cost of writing precise formal specifications. While recent large language models (LLMs) have demonstrated impressive capabilities in theorem proving and verified code generation, how powe…
Test-time scaling has emerged as a powerful paradigm for enhancing the reasoning capabilities of large language models (LLMs) by allocating additional computational resources during inference. However, this paradigm is inherently inefficient due to the generation of redundant and repetitive reasonin…
Large Reasoning Models (LRMs) demonstrate exceptional capability in tackling complex mathematical, logical, and coding tasks by leveraging extended Chain-of-Thought (CoT) reasoning. Test-time scaling methods—such as prolonging CoT with explicit token-level exploration—can push LRMs’ accuracy boundar…
Attention patterns play a crucial role in both training and inference of large language models (LLMs). Prior works have identified individual patterns—such as retrieval heads, sink heads, and diagonal traces—but these observations remain fragmented and lack a unifying explanation. To bridge this gap…
Visual uncertainties such as occlusions, lack of texture, and noise present significant challenges in obtaining accurate kinematic models for safe robotic manipulation. We introduce a probabilistic real-time approach that leverages the human hand as a prior to mitigate these uncertainties. By tracki
Key-Value (KV) cache eviction---which retains the KV pairs of the most important tokens while discarding less important ones---is a critical technique for optimizing both memory usage and inference latency in large language models (LLMs). However, existing approaches often rely on simple heuristics-…
With the development of large language models (LLMs), efficient inference through Key-Value (KV) cache compression has attracted considerable attention, especially for long-context generation. To compress the KV cache, recent methods identify critical KV tokens through static modeling of attention s…
Implementing Boolean functions with circuits consisting of logic gates is fundamental in digital computer design. However, the implemented circuit must be exactly equivalent, which hinders generative neural approaches on this task due to their occasionally wrong predictions. In this study, we introd…
Recent studies have achieved remarkable results for action recognition with human skeletal data by utilizing graph convolutional models. Traditional approaches typically aggregate local spatio-temporal information bottom-up to form a single spatio-temporal global understanding. However, this method…
Point cloud sequence-based 3D action recognition has achieved impressive performance and efficiency. However, existing point cloud sequence modeling methods cannot adequately balance the precision of limb micro-movements with the integrity of posture macro-structure, leading to the loss of crucial i…
KV cache quantization can improve Large Language Models (LLMs) inference throughput and latency in long contexts and large batch-size scenarios while preserving LLMs effectiveness. However, current methods have three unsolved issues: overlooking layer-wise sensitivity to KV cache quantization, high…
Recently, convolutional neural networks for finger vein recognition have gained attention, but their application in IoT smart home security is underexplored. Existing methods typically require networks to identify all categories in a dataset, leading to high parameter demands, which is inefficient g…
Medical image reconstruction from measurement data is a vital but challenging inverse problem. Deep learning approaches have achieved promising results, but often requires paired measurement and high-quality images, which is typically simulated through a forward model, i.e., retrospective reconstruc…
Logic Synthesis (LS) plays a vital role in chip design. A key task in LS is to simplify circuits---modeled by directed acyclic graphs (DAGs)---with functionality-equivalent transformations. To tackle this task, many LS heuristics apply transformations to subgraphs---rooted at each node on an input D…
This letter proposes an adaptive neural network synchronous tracking control strategy that can be suitable for event-triggered mechanism in response to the modeling uncertainties and communication delays in bilateral teleoperation systems. Through introducing the event-triggered mechanism with the a
Shape formation of robot swarms is quite challenging when the robot number varies, as the number often needs to match the number of goal locations in the shape. For this challenge, state-of-the-art methods characterize the shape as a continuous region and distribute robots in the region. However, su
The methods of shape formation in robot swarms are usually classified into two categories by whether assignment is used or not. The first is to use target assignment to assemble precise formation. However, the additional algorithm for re-assignment is required to handle unreasonable situations, whic
We present StreamVC, a streaming voice conversion solution that preserves the content and prosody of any source speech while matching the voice timbre from any target speech. Unlike previous approaches, StreamVC produces the resulting waveform at low latency from the input signal even on a mobile pl…
Logic Synthesis (LS) aims to generate an optimized logic circuit satisfying a given functionality, which generally consists of circuit translation and optimization. It is a challenging and fundamental combinatorial optimization problem in integrated circuit design. Traditional LS approaches rely on…
We introduce a Learning from Demonstration (LID) approach for contact-rich manipulation tasks, i.e., tasks in which the manipulandum's motion is constrained by contact with the environment. Our approach is motivated by the insight that even a large number of demonstrations will often not contain suf…
Payload capacity is a crucial factor for climbing robots, as it directly affects their ability to carry and transport heavy loads during various climbing tasks. However, many dry adhesion-based legged robots prioritize foot design from a bionic perspective to accomplish various climbing tasks while
Recently, many deep stereo matching methods have begun to focus on cross-domain performance, achieving impressive achievements. However, these methods did not deal with the significant volatility of generalization performance among different training epochs. Inspired by masked representation learnin…
We present a novel learning from demonstration approach which uses environmental constraints as the underlying representation to interpret and reproduce demonstrations. This representation based on environmental constraints separates the information that facilitates generalization from the informati
General-purpose planning algorithms for automated driving combine mission, behavior, and local motion planning. Such planning algorithms map features of the environment and driving kinematics into complex reward functions. To achieve this, planning experts often rely on linear reward functions. The…
General-purpose trajectory planning algorithms for automated driving utilize complex reward functions to perform a combined optimization of strategic, behavioral, and kinematic features. The specification and tuning of a single reward function is a tedious task and does not generalize over a large s…
This paper presents an FPGA-based prototype of the dual link algorithm that maximize the achievable weighted sum rate for MIMO interference network. The iterative algorithm is fast monotone convergent but it must be completed quickly through pilot signaling. Therefore we propose an FPGA-based implem…