We revisit and slightly modify the proof of the Gaussian Hanson-Wright inequality where we keep track of the absolute constant in its formulation.
相關內容
The complexity of the alignment problem stems from the fact that existing methods are unstable. Researchers continuously invent various tricks to address this shortcoming. For instance, in the fundamental Reinforcement Learning From Human Feedback (RLHF) technique of Language Model alignment, in addition to reward maximization, the Kullback-Leibler divergence between the trainable policy and the SFT policy is minimized. This addition prevents the model from being overfitted to the Reward Model (RM) and generating texts that are out-of-domain for the RM. The Direct Preference Optimization (DPO) method reformulates the optimization task of RLHF and eliminates the Reward Model while tacitly maintaining the requirement for the policy to be close to the SFT policy. In our paper, we argue that this implicit limitation in the DPO method leads to sub-optimal results. We propose a new method called Trust Region DPO (TR-DPO), which updates the reference policy during training. With such a straightforward update, we demonstrate the effectiveness of TR-DPO against DPO on the Anthropic HH and TLDR datasets. We show that TR-DPO outperforms DPO by up to 19%, measured by automatic evaluation with GPT-4. The new alignment approach that we propose allows us to improve the quality of models across several parameters at once, such as coherence, correctness, level of detail, helpfulness, and harmlessness.
Though multimodal emotion recognition has achieved significant progress over recent years, the potential of rich synergic relationships across the modalities is not fully exploited. In this paper, we introduce Recursive Joint Cross-Modal Attention (RJCMA) to effectively capture both intra- and inter-modal relationships across audio, visual, and text modalities for dimensional emotion recognition. In particular, we compute the attention weights based on cross-correlation between the joint audio-visual-text feature representations and the feature representations of individual modalities to simultaneously capture intra- and intermodal relationships across the modalities. The attended features of the individual modalities are again fed as input to the fusion model in a recursive mechanism to obtain more refined feature representations. We have also explored Temporal Convolutional Networks (TCNs) to improve the temporal modeling of the feature representations of individual modalities. Extensive experiments are conducted to evaluate the performance of the proposed fusion model on the challenging Affwild2 dataset. By effectively capturing the synergic intra- and inter-modal relationships across audio, visual, and text modalities, the proposed fusion model achieves a Concordance Correlation Coefficient (CCC) of 0.585 (0.542) and 0.674 (0.619) for valence and arousal respectively on the validation set(test set). This shows a significant improvement over the baseline of 0.240 (0.211) and 0.200 (0.191) for valence and arousal, respectively, in the validation set (test set), achieving second place in the valence-arousal challenge of the 6th Affective Behavior Analysis in-the-Wild (ABAW) competition.
The stochastic finite volume method offers an efficient one-pass approach for assessing uncertainty in hyperbolic conservation laws. Still, it struggles with the curse of dimensionality when dealing with multiple stochastic variables. We introduce the stochastic finite volume method within the tensor-train framework to counteract this limitation. This integration, however, comes with its own set of difficulties, mainly due to the propensity for shock formation in hyperbolic systems. To overcome these issues, we have developed a tensor-train-adapted stochastic finite volume method that employs a global WENO reconstruction, making it suitable for such complex systems. This approach represents the first step in designing tensor-train techniques for hyperbolic systems and conservation laws involving shocks.
We propose a method to guide Large Language Models (LLMs) in generating structured content adhering to specific conventions without fine-tuning. By utilizing coroutine-based content generation constraints through a pre-agreed context-free grammar (CFG), LLMs are directed during decoding to produce formal language compliant outputs. This enhances stability and consistency in generating target data structures, types, or instructions, reducing application development complexities. Experimentally, error rates of GPT-2 and Gemma exceed 95% for DSLs longer than 36 and 282 tokens, respectively. We introduce YieldLang, a coroutine-based DSL generation framework, and evaluate it with LLMs on various tasks including JSON and Mermaid flowchart generation. Compared to benchmarks, our approach improves accuracy by 1.09 to 11.6 times, with LLMs requiring only about 16.5% of the samples to generate JSON effectively. This enhances usability of LLM-generated content for computer programs.
We propose a method to guide Large Language Models (LLMs) in generating structured content adhering to specific conventions without fine-tuning. By utilizing coroutine-based content generation constraints through a pre-agreed context-free grammar (CFG), LLMs are directed during decoding to produce formal language compliant outputs. This enhances stability and consistency in generating target data structures, types, or instructions, reducing application development complexities. Experimentally, error rates of GPT-2 and Gemma exceed 95% for DSLs longer than 36 and 282 tokens, respectively. We introduce YieldLang, a coroutine-based DSL generation framework, and evaluate it with LLMs on various tasks including JSON and Mermaid flowchart generation. Compared to benchmarks, our approach improves accuracy by 1.09 to 11.6 times, with LLMs requiring only about 16.5% of the samples to generate JSON effectively. This enhances usability of LLM-generated content for computer programs.
Aspect Category Detection (ACD) aims to identify implicit and explicit aspects in a given review sentence. The state-of-the-art approaches for ACD use Deep Neural Networks (DNNs) to address the problem as a multi-label classification task. However, learning category-specific representations heavily rely on the amount of labeled examples, which may not readily available in real-world scenarios. In this paper, we propose a novel approach to tackle the ACD task by combining DNNs with Gradual Machine Learning (GML) in a supervised setting. we aim to leverage the strength of DNN in semantic relation modeling, which can facilitate effective knowledge transfer between labeled and unlabeled instances during the gradual inference of GML. To achieve this, we first analyze the learned latent space of the DNN to model the relations, i.e., similar or opposite, between instances. We then represent these relations as binary features in a factor graph to efficiently convey knowledge. Finally, we conduct a comparative study of our proposed solution on real benchmark datasets and demonstrate that the GML approach, in collaboration with DNNs for feature extraction, consistently outperforms pure DNN solutions.
We study evaluating a policy under best- and worst-case perturbations to a Markov decision process (MDP), given transition observations from the original MDP, whether under the same or different policy. This is an important problem when there is the possibility of a shift between historical and future environments, due to e.g. unmeasured confounding, distributional shift, or an adversarial environment. We propose a perturbation model that can modify transition kernel densities up to a given multiplicative factor or its reciprocal, which extends the classic marginal sensitivity model (MSM) for single time step decision making to infinite-horizon RL. We characterize the sharp bounds on policy value under this model, that is, the tightest possible bounds given by the transition observations from the original MDP, and we study the estimation of these bounds from such transition observations. We develop an estimator with several appealing guarantees: it is semiparametrically efficient, and remains so even when certain necessary nuisance functions such as worst-case Q-functions are estimated at slow nonparametric rates. It is also asymptotically normal, enabling easy statistical inference using Wald confidence intervals. In addition, when certain nuisances are estimated inconsistently we still estimate a valid, albeit possibly not sharp bounds on the policy value. We validate these properties in numeric simulations. The combination of accounting for environment shifts from train to test (robustness), being insensitive to nuisance-function estimation (orthogonality), and accounting for having only finite samples to learn from (inference) together leads to credible and reliable policy evaluation.
Emotion Recognition in Conversation (ERC) involves detecting the underlying emotion behind each utterance within a conversation. Effectively generating representations for utterances remains a significant challenge in this task. Recent works propose various models to address this issue, but they still struggle with differentiating similar emotions such as excitement and happiness. To alleviate this problem, We propose an Emotion-Anchored Contrastive Learning (EACL) framework that can generate more distinguishable utterance representations for similar emotions. To achieve this, we utilize label encodings as anchors to guide the learning of utterance representations and design an auxiliary loss to ensure the effective separation of anchors for similar emotions. Moreover, an additional adaptation process is proposed to adapt anchors to serve as effective classifiers to improve classification performance. Across extensive experiments, our proposed EACL achieves state-of-the-art emotion recognition performance and exhibits superior performance on similar emotions. Our code is available at //github.com/Yu-Fangxu/EACL.
Emotion recognition in conversation (ERC) aims to detect the emotion label for each utterance. Motivated by recent studies which have proven that feeding training examples in a meaningful order rather than considering them randomly can boost the performance of models, we propose an ERC-oriented hybrid curriculum learning framework. Our framework consists of two curricula: (1) conversation-level curriculum (CC); and (2) utterance-level curriculum (UC). In CC, we construct a difficulty measurer based on "emotion shift" frequency within a conversation, then the conversations are scheduled in an "easy to hard" schema according to the difficulty score returned by the difficulty measurer. For UC, it is implemented from an emotion-similarity perspective, which progressively strengthens the model's ability in identifying the confusing emotions. With the proposed model-agnostic hybrid curriculum learning strategy, we observe significant performance boosts over a wide range of existing ERC models and we are able to achieve new state-of-the-art results on four public ERC datasets.
Graph Neural Networks (GNNs) have recently become increasingly popular due to their ability to learn complex systems of relations or interactions arising in a broad spectrum of problems ranging from biology and particle physics to social networks and recommendation systems. Despite the plethora of different models for deep learning on graphs, few approaches have been proposed thus far for dealing with graphs that present some sort of dynamic nature (e.g. evolving features or connectivity over time). In this paper, we present Temporal Graph Networks (TGNs), a generic, efficient framework for deep learning on dynamic graphs represented as sequences of timed events. Thanks to a novel combination of memory modules and graph-based operators, TGNs are able to significantly outperform previous approaches being at the same time more computationally efficient. We furthermore show that several previous models for learning on dynamic graphs can be cast as specific instances of our framework. We perform a detailed ablation study of different components of our framework and devise the best configuration that achieves state-of-the-art performance on several transductive and inductive prediction tasks for dynamic graphs.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191