Recent advancements in offline Reinforcement Learning (Offline RL) have led to an increased focus on methods based on conservative policy updates to address the Out-of-Distribution (OOD) issue. These methods typically involve adding behavior regularization or modifying the critic learning objective, focusing primarily on states or actions with substantial dataset support. However, we challenge this prevailing notion by asserting that the absence of an action or state from a dataset does not necessarily imply its suboptimality. In this paper, we propose a novel approach to tackle the OOD problem. We introduce an offline RL teacher-student framework, complemented by a policy similarity measure. This framework enables the student policy to gain insights not only from the offline RL dataset but also from the knowledge transferred by a teacher policy. The teacher policy is trained using another dataset consisting of state-action pairs, which can be viewed as practical domain knowledge acquired without direct interaction with the environment. We believe this additional knowledge is key to effectively solving the OOD issue. This research represents a significant advancement in integrating a teacher-student network into the actor-critic framework, opening new avenues for studies on knowledge transfer in offline RL and effectively addressing the OOD challenge.
Wetlands are important to communities, offering benefits ranging from water purification, and flood protection to recreation and tourism. Therefore, identifying and prioritizing potential wetland areas is a critical decision problem. While data-driven solutions are feasible, this is complicated by significant data sparsity due to the low proportion of wetlands (3-6\%) in many areas of interest in the southwestern US. This makes it hard to develop data-driven models that can help guide the identification of additional wetland areas. To solve this limitation, we propose two strategies: (1) The first of these is knowledge transfer from regions with rich wetlands (such as the Eastern US) to sparser regions (such as the Southwestern area with few wetlands). Recognizing that these regions are likely to be very different from each other in terms of soil characteristics, population distribution, and land use, we propose a domain disentanglement strategy that identifies and transfers only the applicable aspects of the learned model. (2) We complement this with a spatial data enrichment strategy that relies on an adaptive propagation mechanism. This mechanism differentiates between node pairs that have positive and negative impacts on each other for Graph Neural Networks (GNNs). To summarize, given two spatial cells belonging to different regions, we identify domain-specific and domain-shareable features, and, for each region, we rely on adaptive propagation to enrich features with the features of surrounding cells. We conduct rigorous experiments to substantiate our proposed method's effectiveness, robustness, and scalability compared to state-of-the-art baselines. Additionally, an ablation study demonstrates that each module is essential in prioritizing potential wetlands, which justifies our assumption.
Gene expression programming is an evolutionary optimization algorithm with the potential to generate interpretable and easily implementable equations for regression problems. Despite knowledge gained from previous optimizations being potentially available, the initial candidate solutions are typically generated randomly at the beginning and often only include features or terms based on preliminary user assumptions. This random initial guess, which lacks constraints on the search space, typically results in higher computational costs in the search for an optimal solution. Meanwhile, transfer learning, a technique to reuse parts of trained models, has been successfully applied to neural networks. However, no generalized strategy for its use exists for symbolic regression in the context of evolutionary algorithms. In this work, we propose an approach for integrating transfer learning with gene expression programming applied to symbolic regression. The constructed framework integrates Natural Language Processing techniques to discern correlations and recurring patterns from equations explored during previous optimizations. This integration facilitates the transfer of acquired knowledge from similar tasks to new ones. Through empirical evaluation of the extended framework across a range of univariate problems from an open database and from the field of computational fluid dynamics, our results affirm that initial solutions derived via a transfer learning mechanism enhance the algorithm's convergence rate towards improved solutions.
Despite advancements in large language models (LLMs), non-factual responses remain prevalent. Unlike extensive studies on post-hoc detection of such responses, this work studies non-factuality prediction (NFP), aiming to predict whether an LLM will generate a non-factual response to a question before the generation process. Previous efforts on NFP have demonstrated LLMs' awareness of their internal knowledge, but they still face challenges in efficiency and transferability. In this work, we propose a lightweight NFP model named Factuality Lens (FacLens), which effectively probes hidden representations of questions for the NFP task. Besides, we discover that hidden question representations sourced from different LLMs exhibit similar NFP patterns, which enables the transferability of FacLens across LLMs to reduce development costs. Extensive experiments highlight FacLens's superiority in both effectiveness and efficiency.
Visually detecting camouflaged objects is a hard problem for both humans and computer vision algorithms. Strong similarities between object and background appearance make the task significantly more challenging than traditional object detection or segmentation tasks. Current state-of-the-art models use either convolutional neural networks or vision transformers as feature extractors. They are trained in a fully supervised manner and thus need a large amount of labeled training data. In this paper, both self-supervised and frugal learning methods are introduced to the task of Camouflaged Object Detection (COD). The overall goal is to fine-tune two COD reference methods, namely SINet-V2 and HitNet, pre-trained for camouflaged animal detection to the task of camouflaged human detection. Therefore, we use the public dataset CPD1K that contains camouflaged humans in a forest environment. We create a strong baseline using supervised frugal transfer learning for the fine-tuning task. Then, we analyze three pseudo-labeling approaches to perform the fine-tuning task in a self-supervised manner. Our experiments show that we achieve similar performance by pure self-supervision compared to fully supervised frugal learning.
Policy Distillation (PD) has become an effective method to improve deep reinforcement learning tasks. The core idea of PD is to distill policy knowledge from a teacher agent to a student agent. However, the teacher-student framework requires a well-trained teacher model which is computationally expensive.In the light of online knowledge distillation, we study the knowledge transfer between different policies that can learn diverse knowledge from the same environment.In this work, we propose Online Policy Distillation (OPD) with Decision-Attention (DA), an online learning framework in which different policies operate in the same environment to learn different perspectives of the environment and transfer knowledge to each other to obtain better performance together. With the absence of a well-performance teacher policy, the group-derived targets play a key role in transferring group knowledge to each student policy. However, naive aggregation functions tend to cause student policies quickly homogenize. To address the challenge, we introduce the Decision-Attention module to the online policies distillation framework. The Decision-Attention module can generate a distinct set of weights for each policy to measure the importance of group members. We use the Atari platform for experiments with various reinforcement learning algorithms, including PPO and DQN. In different tasks, our method can perform better than an independent training policy on both PPO and DQN algorithms. This suggests that our OPD-DA can transfer knowledge between different policies well and help agents obtain more rewards.
Multiple Sequence Alignment (MSA) plays a pivotal role in unveiling the evolutionary trajectories of protein families. The accuracy of protein structure predictions is often compromised for protein sequences that lack sufficient homologous information to construct high quality MSA. Although various methods have been proposed to generate virtual MSA under these conditions, they fall short in comprehensively capturing the intricate coevolutionary patterns within MSA or require guidance from external oracle models. Here we introduce MSAGPT, a novel approach to prompt protein structure predictions via MSA generative pretraining in the low MSA regime. MSAGPT employs a simple yet effective 2D evolutionary positional encoding scheme to model complex evolutionary patterns. Endowed by this, its flexible 1D MSA decoding framework facilitates zero or few shot learning. Moreover, we demonstrate that leveraging the feedback from AlphaFold2 can further enhance the model capacity via Rejective Fine tuning (RFT) and Reinforcement Learning from AF2 Feedback (RLAF). Extensive experiments confirm the efficacy of MSAGPT in generating faithful virtual MSA to enhance the structure prediction accuracy. The transfer learning capabilities also highlight its great potential for facilitating other protein tasks.
This paper explores the concept formation and alignment within the realm of language models (LMs). We propose a mechanism for identifying concepts and their hierarchical organization within the semantic representations learned by various LMs, encompassing a spectrum from early models like Glove to the transformer-based language models like ALBERT and T5. Our approach leverages the inherent structure present in the semantic embeddings generated by these models to extract a taxonomy of concepts and their hierarchical relationships. This investigation sheds light on how LMs develop conceptual understanding and opens doors to further research to improve their ability to reason and leverage real-world knowledge. We further conducted experiments and observed the possibility of isolating these extracted conceptual representations from the reasoning modules of the transformer-based LMs. The observed concept formation along with the isolation of conceptual representations from the reasoning modules can enable targeted token engineering to open the door for potential applications in knowledge transfer, explainable AI, and the development of more modular and conceptually grounded language models.
Computational prediction of the interaction of T cell receptors (TCRs) and their ligands is a grand challenge in immunology. Despite advances in high-throughput assays, specificity-labelled TCR data remains sparse. In other domains, the pre-training of language models on unlabelled data has been successfully used to address data bottlenecks. However, it is unclear how to best pre-train protein language models for TCR specificity prediction. Here we introduce a TCR language model called SCEPTR (Simple Contrastive Embedding of the Primary sequence of T cell Receptors), capable of data-efficient transfer learning. Through our model, we introduce a novel pre-training strategy combining autocontrastive learning and masked-language modelling, which enables SCEPTR to achieve its state-of-the-art performance. In contrast, existing protein language models and a variant of SCEPTR pre-trained without autocontrastive learning are outperformed by sequence alignment-based methods. We anticipate that contrastive learning will be a useful paradigm to decode the rules of TCR specificity.
We provide a statistical analysis of regularization-based continual learning on a sequence of linear regression tasks, with emphasis on how different regularization terms affect the model performance. We first derive the convergence rate for the oracle estimator obtained as if all data were available simultaneously. Next, we consider a family of generalized $\ell_2$-regularization algorithms indexed by matrix-valued hyperparameters, which includes the minimum norm estimator and continual ridge regression as special cases. As more tasks are introduced, we derive an iterative update formula for the estimation error of generalized $\ell_2$-regularized estimators, from which we determine the hyperparameters resulting in the optimal algorithm. Interestingly, the choice of hyperparameters can effectively balance the trade-off between forward and backward knowledge transfer and adjust for data heterogeneity. Moreover, the estimation error of the optimal algorithm is derived explicitly, which is of the same order as that of the oracle estimator. In contrast, our lower bounds for the minimum norm estimator and continual ridge regression show their suboptimality. A byproduct of our theoretical analysis is the equivalence between early stopping and generalized $\ell_2$-regularization in continual learning, which may be of independent interest. Finally, we conduct experiments to complement our theory.
Despite the rapid expansion of smart grids and large volumes of data at the individual consumer level, there are still various cases where adequate data collection to train accurate load forecasting models is challenging or even impossible. This paper proposes adapting an established model-agnostic meta-learning algorithm for short-term load forecasting in the context of few-shot learning. Specifically, the proposed method can rapidly adapt and generalize within any unknown load time series of arbitrary length using only minimal training samples. In this context, the meta-learning model learns an optimal set of initial parameters for a base-level learner recurrent neural network. The proposed model is evaluated using a dataset of historical load consumption data from real-world consumers. Despite the examined load series' short length, it produces accurate forecasts outperforming transfer learning and task-specific machine learning methods by $12.5\%$. To enhance robustness and fairness during model evaluation, a novel metric, mean average log percentage error, is proposed that alleviates the bias introduced by the commonly used MAPE metric. Finally, a series of studies to evaluate the model's robustness under different hyperparameters and time series lengths is also conducted, demonstrating that the proposed approach consistently outperforms all other models.
FDA Medical Device recalls are critical and time-sensitive events, requiring swift identification of impacted devices to inform the public of a recall event and ensure patient safety. The OpenFDA device recall dataset contains valuable information about ongoing device recall actions, but manually extracting relevant device information from the recall action summaries is a time-consuming task. Named Entity Recognition (NER) is a task in Natural Language Processing (NLP) that involves identifying and categorizing named entities in unstructured text. Existing NER models, including domain-specific models like BioBERT, struggle to correctly identify medical device trade names, part numbers and component terms within these summaries. To address this, we propose DeviceBERT, a medical device annotation, pre-processing and enrichment pipeline, which builds on BioBERT to identify and label medical device terminology in the device recall summaries with improved accuracy. Furthermore, we demonstrate that our approach can be applied effectively for performing entity recognition tasks where training data is limited or sparse.
The sample efficiency of Bayesian optimization algorithms depends on carefully crafted acquisition functions (AFs) guiding the sequential collection of function evaluations. The best-performing AF can vary significantly across optimization problems, often requiring ad-hoc and problem-specific choices. This work tackles the challenge of designing novel AFs that perform well across a variety of experimental settings. Based on FunSearch, a recent work using Large Language Models (LLMs) for discovery in mathematical sciences, we propose FunBO, an LLM-based method that can be used to learn new AFs written in computer code by leveraging access to a limited number of evaluations for a set of objective functions. We provide the analytic expression of all discovered AFs and evaluate them on various global optimization benchmarks and hyperparameter optimization tasks. We show how FunBO identifies AFs that generalize well in and out of the training distribution of functions, thus outperforming established general-purpose AFs and achieving competitive performance against AFs that are customized to specific function types and are learned via transfer-learning algorithms.
Cross-lingual summarization (XLS) aims to generate a summary in a target language different from the source language document. While large language models (LLMs) have shown promising zero-shot XLS performance, their few-shot capabilities on this task remain unexplored, especially for low-resource languages with limited parallel data. In this paper, we investigate the few-shot XLS performance of various models, including Mistral-7B-Instruct-v0.2, GPT-3.5, and GPT-4. Our experiments demonstrate that few-shot learning significantly improves the XLS performance of LLMs, particularly GPT-3.5 and GPT-4, in low-resource settings. However, the open-source model Mistral-7B-Instruct-v0.2 struggles to adapt effectively to the XLS task with limited examples. Our findings highlight the potential of few-shot learning for improving XLS performance and the need for further research in designing LLM architectures and pre-training objectives tailored for this task. We provide a future work direction to explore more effective few-shot learning strategies and to investigate the transfer learning capabilities of LLMs for cross-lingual summarization.
Large Language Models (LLMs) have shown exceptional performance in text processing. Notably, LLMs can synthesize information from large datasets and explain their decisions similarly to human reasoning through a chain of thought (CoT). An emerging application of LLMs is the handling and interpreting of numerical data, where fine-tuning enhances their performance over basic inference methods. This paper proposes a novel approach to training LLMs using knowledge transfer from a random forest (RF) ensemble, leveraging its efficiency and accuracy. By converting RF decision paths into natural language statements, we generate outputs for LLM fine-tuning, enhancing the model's ability to classify and explain its decisions. Our method includes verifying these rules through established classification metrics, ensuring their correctness. We also examine the impact of preprocessing techniques on the representation of numerical data and their influence on classification accuracy and rule correctness
Category discovery methods aim to find novel categories in unlabeled visual data. At training time, a set of labeled and unlabeled images are provided, where the labels correspond to the categories present in the images. The labeled data provides guidance during training by indicating what types of visual properties and features are relevant for performing discovery in the unlabeled data. As a result, changing the categories present in the labeled set can have a large impact on what is ultimately discovered in the unlabeled set. Despite its importance, the impact of labeled data selection has not been explored in the category discovery literature to date. We show that changing the labeled data can significantly impact discovery performance. Motivated by this, we propose two new approaches for automatically selecting the most suitable labeled data based on the similarity between the labeled and unlabeled data. Our observation is that, unlike in conventional supervised transfer learning, the best labeled is neither too similar, nor too dissimilar, to the unlabeled categories. Our resulting approaches obtains state-of-the-art discovery performance across a range of challenging fine-grained benchmark datasets.
Data often arrives in sequence over time in real-world deep learning applications such as autonomous driving. When new training data is available, training the model from scratch undermines the benefit of leveraging the learned knowledge, leading to significant training costs. Warm-starting from a previously trained checkpoint is the most intuitive way to retain knowledge and advance learning. However, existing literature suggests that this warm-starting degrades generalization. In this paper, we advocate for warm-starting but stepping out of the previous converging point, thus allowing a better adaptation to new data without compromising previous knowledge. We propose Knowledge Consolidation and Acquisition (CKCA), a continuous model improvement algorithm with two novel components. First, a novel feature regularization (FeatReg) to retain and refine knowledge from existing checkpoints; Second, we propose adaptive knowledge distillation (AdaKD), a novel approach to forget mitigation and knowledge transfer. We tested our method on ImageNet using multiple splits of the training data. Our approach achieves up to $8.39\%$ higher top1 accuracy than the vanilla warm-starting and consistently outperforms the prior art with a large margin.
With the booming of Large Language Models (LLMs), prompt-learning has become a promising method mainly researched in various research areas. Recently, many attempts based on prompt-learning have been made to improve the performance of text classification. However, most of these methods are based on heuristic Chain-of-Thought (CoT), and tend to be more complex but less efficient. In this paper, we rethink the LLM-based text classification methodology, propose a simple and effective transfer learning strategy, namely LLMEmbed, to address this classical but challenging task. To illustrate, we first study how to properly extract and fuse the text embeddings via various lightweight LLMs at different network depths to improve their robustness and discrimination, then adapt such embeddings to train the classifier. We perform extensive experiments on publicly available datasets, and the results show that LLMEmbed achieves strong performance while enjoys low training overhead using lightweight LLM backbones compared to recent methods based on larger LLMs, i.e. GPT-3, and sophisticated prompt-based strategies. Our LLMEmbed achieves adequate accuracy on publicly available benchmarks without any fine-tuning while merely use 4% model parameters, 1.8% electricity consumption and 1.5% runtime compared to its counterparts. Code is available at: https://github.com/ChunLiu-cs/LLMEmbed-ACL2024.
Due to its cloud-penetrating capability and independence from solar illumination, satellite Synthetic Aperture Radar (SAR) is the preferred data source for large-scale flood mapping, providing global coverage and including various land cover classes. However, most studies on large-scale SAR-derived flood mapping using deep learning algorithms have primarily focused on flooded open areas, utilizing available open-access datasets (e.g., Sen1Floods11) and with limited attention to urban floods. To address this gap, we introduce \textbf{UrbanSARFloods}, a floodwater dataset featuring pre-processed Sentinel-1 intensity data and interferometric coherence imagery acquired before and during flood events. It contains 8,879 $512\times 512$ chips covering 807,500 $km^2$ across 20 land cover classes and 5 continents, spanning 18 flood events. We used UrbanSARFloods to benchmark existing state-of-the-art convolutional neural networks (CNNs) for segmenting open and urban flood areas. Our findings indicate that prevalent approaches, including the Weighted Cross-Entropy (WCE) loss and the application of transfer learning with pretrained models, fall short in overcoming the obstacles posed by imbalanced data and the constraints of a small training dataset. Urban flood detection remains challenging. Future research should explore strategies for addressing imbalanced data challenges and investigate transfer learning's potential for SAR-based large-scale flood mapping. Besides, expanding this dataset to include additional flood events holds promise for enhancing its utility and contributing to advancements in flood mapping techniques.
Contrastive language-audio pre-training (CLAP) enables zero-shot (ZS) inference of audio and exhibits promising performance in several classification tasks. However, conventional audio representations are still crucial for many tasks where ZS is not applicable (e.g., regression problems). Here, we explore a new representation, a general-purpose audio-language representation, that performs well in both ZS and transfer learning. To do so, we propose a new method, M2D-CLAP, which combines self-supervised learning Masked Modeling Duo (M2D) and CLAP. M2D learns an effective representation to model audio signals, and CLAP aligns the representation with text embedding. As a result, M2D-CLAP learns a versatile representation that allows for both ZS and transfer learning. Experiments show that M2D-CLAP performs well on linear evaluation, fine-tuning, and ZS classification with a GTZAN state-of-the-art of 75.17%, thus achieving a general-purpose audio-language representation.