TimeZero: Temporal Video Grounding with Reasoning-Guided LVLM

We introduce TimeZero, a reasoning-guided LVLM designed for the temporal video grounding (TVG) task. This task requires precisely localizing relevant video segments within long videos based on a given language query. TimeZero tackles this challenge by extending the inference process, enabling the model to reason about video-language relationships solely through reinforcement learning. To evaluate the effectiveness of TimeZero, we conduct experiments on two benchmarks, where TimeZero achieves state-of-the-art performance on Charades-STA. Code is available at https://github.com/www-Ye/TimeZero.

Natural Humanoid Robot Locomotion with Generative Motion Prior

Natural and lifelike locomotion remains a fundamental challenge for humanoid robots to interact with human society. However, previous methods either neglect motion naturalness or rely on unstable and ambiguous style rewards. In this paper, we propose a novel Generative Motion Prior (GMP) that provides fine-grained motion-level supervision for the task of natural humanoid robot locomotion. To leverage natural human motions, we first employ whole-body motion retargeting to effectively transfer them to the robot. Subsequently, we train a generative model offline to predict future natural reference motions for the robot based on a conditional variational auto-encoder. During policy training, the generative motion prior serves as a frozen online motion generator, delivering precise and comprehensive supervision at the trajectory level, including joint angles and keypoint positions. The generative motion prior significantly enhances training stability and improves interpretability by offering detailed and dense guidance throughout the learning process. Experimental results in both simulation and real-world environments demonstrate that our method achieves superior motion naturalness compared to existing approaches. Project page can be found at https://sites.google.com/view/humanoid-gmp

First Token Probability Guided RAG for Telecom Question Answering

Large Language Models (LLMs) have garnered significant attention for their impressive general-purpose capabilities. For applications requiring intricate domain knowledge, Retrieval-Augmented Generation (RAG) has shown a distinct advantage in incorporating domain-specific information into LLMs. However, existing RAG research has not fully addressed the challenges of Multiple Choice Question Answering (MCQA) in telecommunications, particularly in terms of retrieval quality and mitigating hallucinations. To tackle these challenges, we propose a novel first token probability guided RAG framework. This framework leverages confidence scores to optimize key hyperparameters, such as chunk number and chunk window size, while dynamically adjusting the context. Our method starts by retrieving the most relevant chunks and generates a single token as the potential answer. The probabilities of all options are then normalized to serve as confidence scores, which guide the dynamic adjustment of the context. By iteratively optimizing the hyperparameters based on these confidence scores, we can continuously improve RAG performance. We conducted experiments to validate the effectiveness of our framework, demonstrating its potential to enhance accuracy in domain-specific MCQA tasks.

LV-CadeNet: Long View Feature Convolution-Attention Fusion Encoder-Decoder Network for Clinical MEG Spike Detection

It is widely acknowledged that the epileptic foci can be pinpointed by source localizing interictal epileptic discharges (IEDs) via Magnetoencephalography (MEG). However, manual detection of IEDs, which appear as spikes in MEG data, is extremely labor intensive and requires considerable professional expertise, limiting the broader adoption of MEG technology. Numerous studies have focused on automatic detection of MEG spikes to overcome this challenge, but these efforts often validate their models on synthetic datasets with balanced positive and negative samples. In contrast, clinical MEG data is highly imbalanced, raising doubts on the real-world efficacy of these models. To address this issue, we introduce LV-CadeNet, a Long View feature Convolution-Attention fusion Encoder-Decoder Network, designed for automatic MEG spike detection in real-world clinical scenarios. Beyond addressing the disparity between training data distribution and clinical test data through semi-supervised learning, our approach also mimics human specialists by constructing long view morphological input data. Moreover, we propose an advanced convolution-attention module to extract temporal and spatial features from the input data. LV-CadeNet significantly improves the accuracy of MEG spike detection, boosting it from 42.31\% to 54.88\% on a novel clinical dataset sourced from Sanbo Brain Hospital Capital Medical University. This dataset, characterized by a highly imbalanced distribution of positive and negative samples, accurately represents real-world clinical scenarios.

A Multi-Scale Spatial-Temporal Network for Wireless Video Transmission

Deep joint source-channel coding (DeepJSCC) has shown promise in wireless transmission of text, speech, and images within the realm of semantic communication. However, wireless video transmission presents greater challenges due to the difficulty of extracting and compactly representing both spatial and temporal features, as well as its significant bandwidth and computational resource requirements. In response, we propose a novel video DeepJSCC (VDJSCC) approach to enable end-to-end video transmission over a wireless channel. Our approach involves the design of a multi-scale vision Transformer encoder and decoder to effectively capture spatial-temporal representations over long-term frames. Additionally, we propose a dynamic token selection module to mask less semantically important tokens from spatial or temporal dimensions, allowing for content-adaptive variable-length video coding by adjusting the token keep ratio. Experimental results demonstrate the effectiveness of our VDJSCC approach compared to digital schemes that use separate source and channel codes, as well as other DeepJSCC schemes, in terms of reconstruction quality and bandwidth reduction.

Channel-Adaptive Wireless Image Semantic Transmission with Learnable Prompts

Recent developments in Deep learning based Joint Source-Channel Coding (DeepJSCC) have demonstrated impressive capabilities within wireless semantic communications system. However, existing DeepJSCC methodologies exhibit limited generalization ability across varying channel conditions, necessitating the preparation of multiple models. Optimal performance is only attained when the channel status during testing aligns precisely with the training channel status, which is very inconvenient for real-life applications. In this paper, we introduce a novel DeepJSCC framework, termed Prompt JSCC (PJSCC), which incorporates a learnable prompt to implicitly integrate the physical channel state into the transmission system. Specifically, the Channel State Prompt (CSP) module is devised to generate prompts based on diverse SNR and channel distribution models. Through the interaction of latent image features with channel features derived from the CSP module, the DeepJSCC process dynamically adapts to varying channel conditions without necessitating retraining. Comparative analyses against leading DeepJSCC methodologies and traditional separate coding approaches reveal that the proposed PJSCC achieves optimal image reconstruction performance across different SNR settings and various channel models, as assessed by Peak Signal-to-Noise Ratio (PSNR) and Learning-based Perceptual Image Patch Similarity (LPIPS) metrics. Furthermore, in real-world scenarios, PJSCC shows excellent memory efficiency and scalability, rendering it readily deployable on resource-constrained platforms to facilitate semantic communications.

Invar-RAG: Invariant LLM-aligned Retrieval for Better Generation

Retrieval-augmented generation (RAG) has shown impressive capability in providing reliable answer predictions and addressing hallucination problems. A typical RAG implementation uses powerful retrieval models to extract external information and large language models (LLMs) to generate answers. In contrast, recent LLM-based retrieval has gained attention for its substantial improvements in information retrieval (IR) due to the LLMs' semantic understanding capability. However, directly applying LLM to RAG systems presents challenges. This may cause feature locality problems as massive parametric knowledge can hinder effective usage of global information across the corpus; for example, an LLM-based retriever often inputs document summaries instead of full documents. Moreover, various pre-trained tasks in LLMs introduce variance, further weakening performance as a retriever. To address these issues, we propose a novel two-stage fine-tuning architecture called Invar-RAG. In the retrieval stage, an LLM-based retriever is constructed by integrating LoRA-based representation learning to tackle feature locality issues. To enhance retrieval performance, we develop two patterns (invariant and variant patterns) and an invariance loss to reduce LLM variance. In the generation stage, a refined fine-tuning method is employed to improve LLM accuracy in generating answers based on retrieved information. Experimental results show that Invar-RAG significantly outperforms existing baselines across three open-domain question answering (ODQA) datasets. Code is available in the Supplementary Material for reproducibility.

Towards Cross-Modal Text-Molecule Retrieval with Better Modality Alignment

Cross-modal text-molecule retrieval model aims to learn a shared feature space of the text and molecule modalities for accurate similarity calculation, which facilitates the rapid screening of molecules with specific properties and activities in drug design. However, previous works have two main defects. First, they are inadequate in capturing modality-shared features considering the significant gap between text sequences and molecule graphs. Second, they mainly rely on contrastive learning and adversarial training for cross-modality alignment, both of which mainly focus on the first-order similarity, ignoring the second-order similarity that can capture more structural information in the embedding space. To address these issues, we propose a novel cross-modal text-molecule retrieval model with two-fold improvements. Specifically, on the top of two modality-specific encoders, we stack a memory bank based feature projector that contain learnable memory vectors to extract modality-shared features better. More importantly, during the model training, we calculate four kinds of similarity distributions (text-to-text, text-to-molecule, molecule-to-molecule, and molecule-to-text similarity distributions) for each instance, and then minimize the distance between these similarity distributions (namely second-order similarity losses) to enhance cross-modal alignment. Experimental results and analysis strongly demonstrate the effectiveness of our model. Particularly, our model achieves SOTA performance, outperforming the previously-reported best result by 6.4%.

On the Crucial Role of Initialization for Matrix Factorization

This work revisits the classical low-rank matrix factorization problem and unveils the critical role of initialization in shaping convergence rates for such nonconvex and nonsmooth optimization. We introduce Nystrom initialization, which significantly improves the global convergence of Scaled Gradient Descent (ScaledGD) in both symmetric and asymmetric matrix factorization tasks. Specifically, we prove that ScaledGD with Nystrom initialization achieves quadratic convergence in cases where only linear rates were previously known. Furthermore, we extend this initialization to low-rank adapters (LoRA) commonly used for finetuning foundation models. Our approach, NoRA, i.e., LoRA with Nystrom initialization, demonstrates superior performance across various downstream tasks and model scales, from 1B to 7B parameters, in large language and diffusion models.

Implicit Regularization of Sharpness-Aware Minimization for Scale-Invariant Problems

Sharpness-aware minimization (SAM) improves generalization of various deep learning tasks. Motivated by popular architectures such as LoRA, we explore the implicit regularization of SAM for scale-invariant problems involving two groups of variables. Instead of focusing on commonly used sharpness, this work introduces a concept termed balancedness, defined as the difference between the squared norm of two variables. This allows us to depict richer global behaviors of SAM. In particular, our theoretical and empirical findings reveal that i) SAM promotes balancedness; and ii) the regularization on balancedness is data-responsive -- outliers have stronger impact. The latter coincides with empirical observations that SAM outperforms SGD in the presence of outliers. Leveraging the implicit regularization, we develop a resource-efficient SAM variant, balancedness-aware regularization (BAR), tailored for scale-invariant problems such as finetuning language models with LoRA. BAR saves 95% computational overhead of SAM, with enhanced test performance across various tasks on RoBERTa, GPT2, and OPT-1.3B.