Hyperparathyroidism (HPT) and thyroid nodule (TN) are caused by abnormalities in the parath/yroid and thyroid glands, respectively. Due to their proximity, small size, and similar ultrasound characteristics, the traditional object detection algorithms often struggle to accurately differentiate between HPT and TN when both lesions coexist in ultrasound images, leading to a high rate of misdiagnosis. In order to achieve accurate detection of HPT and TN when the two lesions coexist, we constructed three comprehensive object detection datasets: one containing only hyperparathyroidism (HPTD), one containing only thyroid nodules (TND), and one mixed dataset that includes both types of lesions (HPT-TND). A novel multi-branch feature fusion DETR network (MB-DETR) is proposed based on the Real-Time Detection Transformer (RT-DETR) model. We redesigned the feature fusion module and incorporated asymmetric convolution to enhance feature extraction. To validate the proposed MB-DETR performance, the experiments have been carried out on the three datasets. Our model achieved a superior performance compared to the state-of-the-art object detection models in the key metrics such as F1, Precision, and Recall, while significantly reducing computational costs. Additionally, the ablation studies confirmed the effectiveness of asymmetric convolution and the multi-branch feature fusion module in terms of enhancement of detection performance. The experimental results show that the Multi-Branch Feature Fusion incorporated with the asymmetric convolution improves the local feature extraction capability of the DETR model. It is concluded that the proposed MB-DETR model outperforms the existing ones in the detection of TN and HPT when both lesions coexist and thus effectively assists in the diagnosis of the correlated disease.

We present Sparse Interpolated Mixture-of-Experts (SIMoE) instruction-tuning, an endto-end algorithm designed to fine-tune a dense pre-trained Large Language Model (LLM) into a MoE-style model that possesses capabilities in multiple specialized domains.During instruction-tuning, SIMoE automatically identifies multiple specialized experts under a specified sparsity constraint, with each expert representing a structurally sparse subset of the seed LLM's parameters that correspond to domainspecific knowledge within the data.SIMoE simultaneously learns an input-dependent expert merging strategy via a router network, leveraging rich cross-expert knowledge for superior downstream generalization that surpasses existing baselines.Empirically, SIMoE consistently achieves state-of-the-art performance on common instruction-tuning benchmarks while maintaining an optimal performance-compute trade-off compared to all baselines.

Recent advances in reasoning with large language models (LLMs) have popularized Long Chain-of-Thought (LCoT), a strategy that encourages deliberate and step-by-step reasoning before producing a final answer.While LCoTs have enabled expert-level performance in complex tasks, how the internal structures of their reasoning chains drive, or even predict, the correctness of final answers remains a critical yet underexplored question.In this work, we present LCoT2Tree, an automated framework that converts sequential LCoTs into hierarchical tree structures and thus enables deeper structural analysis of LLM reasoning.Using graph neural networks (GNNs), we reveal that structural patterns extracted by LCoT2Tree, including exploration, backtracking, and verification, serve as stronger predictors of final performance across a wide range of tasks and models.Leveraging an explainability technique, we further identify critical thought patterns such as over-branching that account for failures.Beyond diagnostic insights, the structural patterns by LCoT2Tree support practical applications, including improving Best-of-N decoding effectiveness.Overall, our results underscore the critical role of internal structures of reasoning chains, positioning LCoT2Tree as a powerful tool for diagnosing, interpreting, and improving reasoning in LLMs.The code is available at our GitHub repository.

The emergence of the foundation model era has sparked immense research interest in utilizing pre-trained representations for continual learning~(CL), yielding a series of strong CL methods with outstanding performance on standard evaluation benchmarks. Nonetheless, there are growing concerns regarding potential data contamination within the massive pre-training datasets. Furthermore, the static nature of standard evaluation benchmarks tends to oversimplify the complexities encountered in real-world CL scenarios, putting CL methods at risk of overfitting to these benchmarks while still lacking robustness needed for more demanding real-world applications. To solve these problems, this paper proposes a general framework to evaluate methods for Continual Learning on Dynamic Benchmarks (CLDyB). CLDyB continuously identifies inherently challenging tasks for the specified CL methods and evolving backbones, and dynamically determines the sequential order of tasks at each time step in CL using a tree-search algorithm, guided by an overarching goal to generate highly challenging task sequences for evaluation. To highlight the significance of dynamic evaluation on the CLDyB, we first simultaneously evaluate multiple state-of-the-art CL methods under CLDyB, resulting in a set of commonly challenging task sequences where existing CL methods tend to underperform. We intend to publicly release these task sequences for the CL community to facilitate the training and evaluation of more robust CL algorithms. Additionally, we perform individual evaluations of the CL methods under CLDyB, yielding informative evaluation results that reveal the specific strengths and weaknesses of each method.

Catastrophic forgetting (CF) poses a significant challenge in machine learning, where a model forgets previously learned information upon learning new tasks. Despite the advanced capabilities of Large Language Models (LLMs), they continue to face challenges with CF during continual learning. The majority of existing research focuses on analyzing forgetting patterns through a singular training sequence, thereby overlooking the intricate effects that diverse tasks have on model behavior. Our study explores CF across various settings, discovering that model forgetting is influenced by both the specific training tasks and the models themselves. To this end, we interpret forgetting by examining the function vector (FV), a compact representation of functions in LLMs, offering a model-dependent indicator for the occurrence of CF. Through theoretical and empirical analyses, we demonstrated that CF in LLMs primarily stems from biases in function activation rather than the overwriting of task processing functions. Leveraging these insights, we propose a novel function vector guided training methodology, incorporating a regularization technique to stabilize the FV and mitigate forgetting. Empirical tests on four benchmarks confirm the effectiveness of our proposed training method, substantiating our theoretical framework concerning CF and model function dynamics.

Continual Learning (CL) with foundation models has recently emerged as a promising paradigm to exploit abundant knowledge acquired during pre-training for tackling sequential tasks. However, existing prompt-based and Low-Rank Adaptation-based (LoRA-based) methods often require expanding a prompt/LoRA pool or retaining samples of previous tasks, which poses significant scalability challenges as the number of tasks grows. To address these limitations, we propose Scalable Decoupled LoRA (SD-LoRA) for class incremental learning, which continually separates the learning of the magnitude and direction of LoRA components without rehearsal. Our empirical and theoretical analysis reveals that SD-LoRA tends to follow a low-loss trajectory and converges to an overlapping low-loss region for all learned tasks, resulting in an excellent stability-plasticity trade-off. Building upon these insights, we introduce two variants of SD-LoRA with further improved parameter efficiency. All parameters of SD-LoRAs can be end-to-end optimized for CL objectives. Meanwhile, they support efficient inference by allowing direct evaluation with the finally trained model, obviating the need for component selection. Extensive experiments across multiple CL benchmarks and foundation models consistently validate the effectiveness of SD-LoRA. The code is available at https://github.com/WuYichen-97/SD-Lora-CL.

Accurately modeling chemical reactions using Artificial Intelligence (AI) can accelerate discovery and development, especially in fields like drug design and material science. Although AI has made remarkable advancements in single molecule recognition, such as predicting molecular properties, the study of interactions between molecules, particularly chemical reactions, has been relatively overlooked. In this paper, we introduce Reaction Graph (RG), a unified graph representation that encapsulates the 3D molecular structures within chemical reactions. RG integrates the molecular graphs of reactants and products into a cohesive framework, effectively capturing the interatomic relationships pertinent to the reaction process. Additionally, it incorporates the 3D structure information of molecules in a simple yet effective manner. We conduct experiments on a range of tasks, including chemical reaction classification, condition prediction, and yield prediction. RG achieves the highest accuracy across six datasets, demonstrating its effectiveness. The code is available at https://github.com/Shadow-Dream/Reaction-Graph.

Low-rank adaptation (LoRA) has emerged as a leading parameter-efficient fine-tuning technique for adapting large foundation models, yet it often locks adapters into suboptimal minima near their initialization. This hampers model generalization and limits downstream operators such as adapter merging and pruning. Here, we propose CoTo, a progressive training strategy that gradually increases adapters’ activation probability over the course of fine-tuning. By stochastically deactivating adapters, CoTo encourages more balanced optimization and broader exploration of the loss landscape. We provide a theoretical analysis showing that CoTo promotes layer-wise dropout stability and linear mode connectivity, and we adopt a cooperative-game approach to quantify each adapter’s marginal contribution. Extensive experiments demonstrate that CoTo consistently boosts single-task performance, enhances multi-task merging accuracy, improves pruning robustness, and reduces training overhead, all while remaining compatible with diverse LoRA variants. Code is available at https://github.com/zwebzone/coto.

The emergence of large language models (LLMs) has opened new opportunities for AI-driven chemical problem solving. However, existing chemical LLMs are typically tailored to specific task formats or narrow domains, limiting their capacity to integrate knowledge and generalize across tasks. Model merging offers a promising route for efficiently combining specialized LLMs into a unified model without access to original training data, which is urgently needed in the chemical domain where in-house data and privacy preservation are critical. However, effective model merging in the chemical domain poses unique challenges: (1) significant disparities among chemical LLMs due to task-specific specialization, and (2) a highly imbalanced distribution of chemical LLMs in targeted downstream tasks, where some are over-benchmarked while others remain underexplored. These challenges intensify model inconsistencies such as parameter interference and accumulated fine-tuning noise, which collectively hinder effective model merging. To this end, we propose Curriculum Model Merging (CMM), a curriculum-based framework that progressively merges expert chemical LLMs in a moderate and continual manner. CMM aims to harmonize their inconsistencies while meantime preserve their domain-specific expertise. Comprehensive experiments on two benchmark datasets show that CMM effectively consolidates task-specific expertise and outperforms the state-of-the-art methods by 29.03\% in terms of overall average performance. Moreover, CMM facilitates chemical knowledge generalization across prediction and generative tasks without sacrificing robustness, exhibiting promising merging performance under both expert-abundant and expert-sparse scenarios.

Model editing has become a *de-facto* practice to address hallucinations and outdated knowledge of large language models (LLMs). However, existing methods are predominantly evaluated in isolation, i.e., one edit at a time, failing to consider a critical scenario of compositional model editing, where multiple edits must be integrated and jointly utilized to answer real-world multifaceted questions. For instance, in medical domains, if one edit informs LLMs that COVID-19 causes "fever" and another that it causes "loss of taste", a qualified compositional editor should enable LLMs to answer the question "What are the symptoms of COVID-19?" with both "fever" and "loss of taste" (and potentially more). In this work, we define and systematically benchmark this compositional model editing (CME) task, identifying three key undesirable issues that existing methods struggle with: *knowledge loss*, *incorrect preceding* and *knowledge sinking*. To overcome these issues, we propose A$^3$E, a novel compositional editor that (1) ***a**daptively combines and **a**daptively regularizes* pre-trained foundation knowledge in LLMs in the stage of edit training and (2) ***a**daptively merges* multiple edits to better meet compositional needs in the stage of edit composing. Extensive experiments demonstrate that A$^3$E improves the composability by at least 22.45\% without sacrificing the performance of non-compositional model editing.

Machine learning-assisted retrosynthesis prediction models have been gaining widespread adoption, though their performances oftentimes degrade significantly when deployed in real-world applications embracing out-of-distribution (OOD) molecules or reactions. Despite steady progress on standard benchmarks, our understanding of existing retrosynthesis prediction models under the premise of distribution shifts remains stagnant. To this end, we first formally sort out two types of distribution shifts in retrosynthesis prediction and construct two groups of benchmark datasets. Next, through comprehensive experiments, we systematically compare state-of-the-art retrosynthesis prediction models on the two groups of benchmarks, revealing the limitations of previous in-distribution evaluation and re-examining the advantages of each model. More remarkably, we are motivated by the above empirical insights to propose two model-agnostic techniques that can improve the OOD generalization of arbitrary off-the-shelf retrosynthesis prediction algorithms. Our preliminary experiments show their high potential with an average performance improvement of 4.6%, and the established benchmarks serve as a foothold for further retrosynthesis prediction research towards OOD generalization.

Tianqi Zhong, Zhaoyi Li, Quan Wang, Linqi Song, Ying Wei, Defu Lian, Zhendong Mao. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Vision-language pretrained models have achieved impressive performance on various downstream tasks. However, their large model sizes hinder their utilization on platforms with limited computational resources. We find that directly using smaller pretrained models and applying magnitude-based pruning on CLIP models leads to in-flexibility and inferior performance. Recent efforts for VLP compression either adopt uni-modal compression metrics resulting in limited performance or involve costly mask-search processes with learnable masks. In this paper, we first propose the Module-wise Pruning Error (MoPE) met-ric, accurately assessing CLIP module importance by performance decline on cross-modal tasks. Using the MoPE metric, we introduce a unified pruning framework applica-ble to both pretraining and task-specific fine-tuning compression stages. For pretraining, MoPE-CLIP effectively leverages knowledge from the teacher model, significantly reducing pretraining costs while maintaining strong zero-shot capabilities. For fine-tuning, consecutive pruning from width to depth yields highly competitive task-specific models. Extensive experiments in two stages demonstrate the effectiveness of the MoPE metric, and MoPE-CLIP outperforms previous state-of-the-art VLP compression methods.

Large Language Models (LLMs) have recently revolutionized the NLP field, while they still fall short in some specific down-stream tasks.In the work, we focus on utilizing LLMs to perform machine translation, where we observe that two patterns of errors frequently occur and drastically affect the translation quality: language mismatch and repetition.The work sets out to explore the potential for mitigating these two issues by leveraging model editing methods, e.g., by locating Feed-Forward Network (FFN) neurons or something that are responsible for the errors and deactivating them in the inference time.We find that directly applying such methods either limited effect on the targeted errors or has significant negative side-effect on the general translation quality, indicating that the located components may also be crucial for ensuring machine translation with LLMs on the rails.To this end, we propose to refine the located components by fetching the intersection of the locating results under different language settings, filtering out the aforementioned information that is irrelevant to targeted errors.The experiment results empirically demonstrate that our methods can effectively reduce the language mismatch and repetition ratios and meanwhile enhance or keep the general translation quality in most cases.

Regularization-based methods have so far been among the *de facto* choices for continual learning. Recent theoretical studies have revealed that these methods all boil down to relying on the Hessian matrix approximation of model weights. However, these methods suffer from suboptimal trade-offs between knowledge transfer and forgetting due to fixed and unchanging Hessian estimations during training. Another seemingly parallel strand of Meta-Continual Learning (Meta-CL) algorithms enforces alignment between gradients of previous tasks and that of the current task. In this work we revisit Meta-CL and for the first time bridge it with regularization-based methods. Concretely, Meta-CL implicitly approximates Hessian in an online manner, which enjoys the benefits of timely adaptation but meantime suffers from high variance induced by random memory buffer sampling. We are thus highly motivated to combine the best of both worlds, through the proposal of Variance Reduced Meta-CL (VR-MCL) to achieve both timely and accurate Hessian approximation. Through comprehensive experiments across three datasets and various settings, we consistently observe that VR-MCL outperforms other SOTA methods, which further validates the effectiveness of VR-MCL.

Retrosynthetic planning is a sequential decision-making process of identifying synthetic routes from the available building block materials to reach a desired target molecule. Though existing planning approaches show promisingly high solving rates and low costs, the trivial route cost evaluation via pre-trained forward reaction prediction models certainly falls short of real-world chemical practice. An alternative option is to annotate the actual cost of a route, such as yield, through chemical experiments or input from chemists, while this often leads to substantial query costs. In order to strike the balance between query costs and route quality evaluation, we propose an Active Retrosynthetic Planning (ARP) framework that remains compatible with the established retrosynthetic planners. On one hand, the proposed ARP trains an actor that decides whether to query the cost of a reaction; on the other hand, it resorts to a critic to estimate the value of a molecule with its preceding reaction cost as input. Those molecules with low reaction costs are preferred to expand first. We apply our framework to different existing approaches on both the benchmark and an expert dataset and demonstrate that it outperforms the existing state-of-the-art approach by 6.2\% in route quality while reducing the query cost by 12.8\%. In addition, ARP consistently plans high-quality routes with either abundant or sparse annotations.

Recent successes suggest that parameter-efficient fine-tuning of foundation models is becoming the state-of-the-art method for transfer learning in vision, gradually replacing the rich literature of alternatives such as meta-learning. In trying to harness the best of both worlds, meta-tuning introduces a subsequent optimization stage of foundation models but has so far only shown limited success and crucially tends to underperform on out-of-distribution (OOD) tasks. In this paper, we introduce Sparse MetA-Tuning (SMAT), a method inspired by sparse mixture-of-experts approaches and trained to isolate subsets of pre-trained parameters automatically for meta-tuning on each task. SMAT successfully overcomes OOD sensitivity and delivers on the promise of enhancing the transfer abilities of vision foundation models beyond parameter-efficient finetuning. We establish new state-of-the-art results on a challenging combination of Meta-Dataset augmented with additional OOD tasks in both zero-shot and gradient-based adaptation settings. In addition, we provide a thorough analysis of the superiority of learned over hand-designed sparsity patterns for sparse expert methods and the pivotal importance of the sparsity level in balancing between in-distribution and out-of-distribution generalization. Our code and models are publicly available.

In Federated Continual Learning (FCL), the challenge lies in effectively facilitating knowledge transfer and enhancing the performance across various tasks on different clients. Current FCL methods predominantly focus on avoiding interference between tasks, thereby overlooking the potential for positive knowledge transfer across tasks learned by different clients at separate time intervals. To address this issue, we introduce a **P**rompt-based kn**ow**le**d**ge transf**er** FCL algorithm, called **Powder**, designed to effectively foster the transfer of knowledge encapsulated in prompts between various sequentially learned tasks and clients. Furthermore, we have devised a unique approach for prompt generation and aggregation, intending to alleviate privacy protection concerns and communication overhead, while still promoting knowledge transfer. Comprehensive experimental results demonstrate the superiority of our method in terms of reduction in communication costs, and enhancement of knowledge transfer. Code is available at https://github.com/piaohongming/Powder.

Catastrophic forgetting remains a core challenge in continual learning (CL), where the models struggle to retain previous knowledge when learning new tasks. While existing replay-based CL methods have been proposed to tackle this challenge by utilizing a memory buffer to store data from previous tasks, they generally overlook the interdependence between previously learned tasks and fail to encapsulate the optimally integrated knowledge in previous tasks, leading to sub-optimal performance of the previous tasks. Against this issue, we first reformulate replay-based CL methods as a unified hierarchical gradient aggregation framework. We then incorporate the Pareto optimization to capture the interrelationship among previously learned tasks and design a Pareto-Optimized CL algorithm (POCL), which effectively enhances the overall performance of past tasks while ensuring the performance of the current task. Comprehensive empirical results demonstrate that the proposed POCL outperforms current state-of-the-art CL methods across multiple datasets and different settings.

To improve meta-generalization, i.e., accommodating out-of-domain meta-testing tasks beyond meta-training ones, is of significance to extending the success of meta-learning beyond standard benchmarks. Previous heterogeneous meta-learning algorithms have shown that tailoring the global meta-knowledge by the learned clusters during meta-training promotes better meta-generalization to novel meta-testing tasks. Inspired by this, we propose a novel multi-objective perspective to sharpen the compositionality of the meta-trained clusters, through which we have empirically validated that the meta-generalization further improves. Grounded on the hierarchically structured meta-learning framework, we formulate a hypervolume loss to evaluate the degree of conflict between multiple cluster-conditioned parameters in the two-dimensional loss space over two randomly chosen tasks belonging to two clusters and two mixed tasks imitating out-of-domain tasks. Experimental results on more than 16 few-shot image classification datasets show not only improved performance on out-of-domain meta-testing datasets but also better clusters in visualization.

Quantization of large language models (LLMs) faces significant challenges, particularly due to the presence of outlier activations that impede efficient low-bit representation. Traditional approaches predominantly address Normal Outliers, which are activations across all tokens with relatively large magnitudes. However, these methods struggle with smoothing Massive Outliers that display significantly larger values, which leads to significant performance degradation in low-bit quantization. In this paper, we introduce DuQuant, a novel approach that utilizes rotation and permutation transformations to more effectively mitigate both massive and normal outliers. First, DuQuant starts by constructing the rotation matrix, using specific outlier dimensions as prior knowledge, to redistribute outliers to adjacent channels by block-wise rotation. Second, We further employ a zigzag permutation to balance the distribution of outliers across blocks, thereby reducing block-wise variance. A subsequent rotation further smooths the activation landscape, enhancing model performance. DuQuant simplifies the quantization process and excels in managing outliers, outperforming the state-of-the-art baselines across various sizes and types of LLMs on multiple tasks, even with 4-bit weight-activation quantization. Our code is available at https://github.com/Hsu1023/DuQuant.

Model editing aims to data-efficiently correct predictive errors of large pre-trained models while ensuring generalization to neighboring failures and locality to minimize unintended effects on unrelated examples. While significant progress has been made in editing Transformer-based large language models, effective strategies for editing vision Transformers (ViTs) in computer vision remain largely untapped. In this paper, we take initial steps towards correcting predictive errors of ViTs, particularly those arising from subpopulation shifts. Taking a locate-then-edit approach, we first address the ``where-to-edit`` challenge by meta-learning a hypernetwork on CutMix-augmented data generated for editing reliability. This trained hypernetwork produces generalizable binary masks that identify a sparse subset of structured model parameters, responsive to real-world failure samples. Afterward, we solve the ``how-to-edit`` problem by simply fine-tuning the identified parameters using a variant of gradient descent to achieve successful edits. To validate our method, we construct an editing benchmark that introduces subpopulation shifts towards natural underrepresented images and AI-generated images, thereby revealing the limitations of pre-trained ViTs for object recognition. Our approach not only achieves superior performance on the proposed benchmark but also allows for adjustable trade-offs between generalization and locality. Our code is available at https://github.com/hustyyq/Where-to-Edit.

Large-scale diffusion models are adept at generating high-fidelity images and facilitating image editing and interpolation. However, they have limitations when tasked with generating images in dynamic, evolving domains. In this paper, we introduce Terra, a novel Time-varying low-rank adapter that offers a fine-tuning framework specifically tailored for domain flow generation. The key innovation of Terra lies in its construction of a continuous parameter manifold through a time variable, with its expressive power analyzed theoretically. This framework not only enables interpolation of image content and style but also offers a generation-based approach to address the domain shift problems in unsupervised domain adaptation and domain generalization. Specifically, Terra transforms images from the source domain to the target domain and generates interpolated domains with various styles to bridge the gap between domains and enhance the model generalization, respectively. We conduct extensive experiments on various benchmark datasets, empirically demonstrate the effectiveness of Terra. Our source code is publicly available on https://github.com/zwebzone/terra.

Unraveling the intricate network of associations among microRNAs (miRNAs), genes, and diseases is pivotal for deciphering molecular mechanisms, refining disease diagnosis, and crafting targeted therapies. Computational strategies, leveraging link prediction within biological graphs, present a cost-efficient alternative to high-cost empirical assays. However, while plenty of methods excel at predicting specific associations, such as miRNA-disease associations (MDAs), miRNA-target interactions (MTIs), and disease-gene associations (DGAs), a holistic approach harnessing diverse data sources for multifaceted association prediction remains largely unexplored. The limited availability of high-quality data, as vitro experiments to comprehensively confirm associations are often expensive and time-consuming, results in a sparse and noisy heterogeneous graph, hindering an accurate prediction of these complex associations. To address this challenge, we propose a novel framework called Global-local aware Heterogeneous Graph Contrastive Learning (GlaHGCL). GlaHGCL combines global and local contrastive learning to improve node embeddings in the heterogeneous graph. In particular, global contrastive learning enhances the robustness of node embeddings against noise by aligning global representations of the original graph and its augmented counterpart. Local contrastive learning enforces representation consistency between functionally similar or connected nodes across diverse data sources, effectively leveraging data heterogeneity and mitigating the issue of data scarcity. The refined node representations are applied to downstream tasks, such as MDA, MTI, and DGA prediction. Experiments show GlaHGCL outperforming state-of-the-art methods, and case studies further demonstrate its ability to accurately uncover new associations among miRNAs, genes, and diseases. We have made the datasets and source code publicly available at https://github.com/Sue-syx/GlaHGCL.

Retrosynthesis aided by artificial intelligence has been a very active and bourgeoning area of research, for its critical role in drug discovery as well as material science. Three categories of solutions, i.e., template-based, template-free, and semi-template methods, constitute mainstream solutions to this problem. In this paper, we focus on template-free methods which are known to be less bothered by the template generalization issue and the atom mapping challenge. Among several remaining problems regarding template-free methods, failing to conform to chemical rules is pronounced. To address the issue, we seek for a pre-training solution to empower the pre-trained model with chemical rules encoded. Concretely, we enforce the atom conservation rule via a molecule reconstruction pre-training task, and the reaction rule that dictates reaction centers via a reaction type guided contrastive pre-training task. In our empirical evaluation, the proposed pre-training solution substantially improves the single-step retrosynthesis accuracies in three downstream datasets.

Despite the rising prevalence of neural sequence models, recent empirical evidences suggest their deficiency in compositional generalization. One of the current de-facto solutions to this problem is compositional data augmentation, aiming to incur additional compositional inductive bias. Nonetheless, the improvement offered by existing handcrafted augmentation strategies is limited when successful systematic generalization of neural sequence models requires multi-grained compositional bias (i.e., not limited to either lexical or structural biases only) or differentiation of training sequences in an imbalanced difficulty distribution. To address the two challenges, we first propose a novel compositional augmentation strategy dubbed Span Substitution (SpanSub) that enables multi-grained composition of substantial substructures in the whole training set. Over and above that, we introduce the Learning to Substitute Span (L2S2) framework which empowers the learning of span substitution probabilities in SpanSub in an end-to-end manner by maximizing the loss of neural sequence models, so as to outweigh those challenging compositions with elusive concepts and novel surroundings. Our empirical results on three standard compositional generalization benchmarks, including SCAN, COGS and GeoQuery (with an improvement of at most 66.5%, 10.3%, 1.2%, respectively), demonstrate the superiority of SpanSub, L2S2 and their combination. © 2023 Association for Computational Linguistics.

Hyperparathyroidism (HPT) is an endocrine disorder characterized by the excessive synthesis and secretion of parathyroid hormone. The excessive parathyroid hormone can cause bone calcium loss, vascular calcification, and cardiovascular diseases. Because of the multiple and reversible distribution of HPT, the accurate localization of the lesion during clinical diagnoses is challenging, often resulting in misdiagnoses or missed diagnoses. The computer-aided diagnosis (CAD) can improve the diagnostic accuracy. The joint utilization of the Squeeze-and-Excitation (SE) attention module and deformable transformer networks can achieve excellent detection performance for HPT. By improving the weight mapping approach within the SE module, we effectively enhanced the joint network’s detection performance for HPT.

We aim at advancing blind image quality assessment (BIQA), which predicts the human perception of image quality without any reference information. We develop a general and automated multitask learning scheme for BIQA to exploit auxiliary knowledge from other tasks, in a way that the model parameter sharing and the loss weighting are determined automatically. Specifically, we first describe all candidate label combinations (from multiple tasks) using a textual template, and compute the joint probability from the cosine similarities of the visual-textual embeddings. Predictions of each task can be inferred from the joint distribution, and optimized by carefully designed loss functions. Through comprehensive experiments on learning three tasks - BIQA, scene classification, and distortion type identification, we verify that the proposed BIQA method 1) benefits from the scene classification and distortion type identification tasks and outperforms the state-of-the-art on multiple IQA datasets, 2) is more robust in the group maximum differentiation competition, and 3) realigns the quality annotations from different IQA datasets more effectively. The source code is available at https://github.com/zwx8981/LIQE.

A multitude of prevalent pre-trained models mark a major milestone in the development of artificial intelligence, while fine-tuning has been a common practice that enables pretrained models to figure prominently in a wide array of target datasets. Our empirical results reveal that off-the-shelf finetuning techniques are far from adequate to mitigate negative transfer caused by two types of underperforming features in a pre-trained model, including rare features and spuriously correlated features. Rooted in structural causal models of predictions after fine-tuning, we propose a Concept-wise fine-tuning (Concept-Tuning) approach which refines feature representations in the level of patches with each patch encoding a concept. Concept-Tuning minimizes the negative impacts of rare features and spuriously correlated features by (1) maximizing the mutual information between examples in the same category with regard to a slice of rare features (a patch) and (2) applying front-door adjustment via attention neural networks in channels and feature slices (patches). The proposed Concept-Tuning consistently and significantly (by up to 4.76%) improves prior state-of-the-art fine-tuning methods on eleven datasets, diverse pre-training strategies (supervised and self-supervised ones), various network architectures, and sample sizes in a target dataset.

The success of meta-learning on out-of-distribution (OOD) tasks in the wild has proved to be hit-and-miss. To safeguard the generalization capability of the meta-learned prior knowledge to OOD tasks, in particularly safety-critical applications, necessitates detection of an OOD task followed by adaptation of the task towards the prior. Nonetheless, the reliability of estimated uncertainty on OOD tasks by existing Bayesian meta-learning methods is restricted by incomplete coverage of the feature distribution shift and insufficient expressiveness of the meta-learned prior. Besides, they struggle to adapt an OOD task, running parallel to the line of cross-domain task adaptation solutions which are vulnerable to overfitting. To this end, we build a single coherent framework that supports both detection and adaptation of OOD tasks, while remaining compatible with off-the-shelf meta-learning backbones. The proposed Energy-Based Meta-Learning (EBML) framework learns to characterize any arbitrary meta-training task distribution with the composition of two expressive neural-network-based energy functions. We deploy the sum of the two energy functions, being proportional to the joint distribution of a task, as a reliable score for detecting OOD tasks; during meta-testing, we adapt the OOD task to in-distribution tasks by energy minimization. Experiments on four regression and classification datasets demonstrate the effectiveness of our proposal.

Compositionality facilitates the comprehension of novel objects using acquired concepts and the maintenance of a knowledge pool. This is particularly crucial for continual learners to prevent catastrophic forgetting and enable compositionally forward transfer of knowledge. However, the existing state-of-the-art benchmarks inadequately evaluate the capability of compositional generalization, leaving an intriguing question unanswered. To comprehensively assess this capability, we introduce two vision benchmarks, namely Compositional GQA (CGQA) and Compositional OBJects365 (COBJ), along with a novel evaluation framework called Compositional Few-Shot Testing (CFST). These benchmarks evaluate the systematicity, productivity, and substitutivity aspects of compositional generalization. Experimental results on five baselines and two modularity-based methods demonstrate that current continual learning techniques do exhibit somewhat favorable compositionality in their learned feature extractors. Nonetheless, further efforts are required in developing modularity-based approaches to enhance compositional generalization. We anticipate that our proposed benchmarks and evaluation protocol will foster research on continual learning and compositionality.

Allergies have become an emerging public health problem worldwide. The most effective way to prevent allergies is to find the causative allergen at the source and avoid re-exposure. However, most of the current computational methods used to identify allergens were based on homology or conventional machine learning methods, which were inefficient and still had room to be improved for the detection of allergens with low homology. In addition, few methods based on deep learning were reported, although deep learning has been successfully applied to several tasks in protein sequence analysis. In the present work, a deep neural network-based model, called DeepAlgPro, was proposed to identify allergens. We showed its great accuracy and applicability to large-scale forecasts by comparing it to other available tools. Additionally, we used ablation experiments to demonstrate the critical importance of the convolutional module in our model. Moreover, further analyses showed that epitope features contributed to model decision-making, thus improving the model's interpretability. Finally, we found that DeepAlgPro was capable of detecting potential new allergens. Overall, DeepAlgPro can serve as powerful software for identifying allergens.

Transferability estimation has been an essential tool in selecting a pre-trained model and the layers in it for transfer learning, to transfer, so as to maximize the performance on a target task and prevent negative transfer. Existing estimation algorithms either require intensive training on target tasks or have difficulties in evaluating the transferability between layers. To this end, we propose a simple, efficient, and effective transferability measure named TransRate. Through a single pass over examples of a target task, TransRate measures the transferability as the mutual information between features of target examples extracted by a pre-trained model and their labels. We overcome the challenge of efficient mutual information estimation by resorting to coding rate that serves as an effective alternative to entropy. From the perspective of feature representation, the resulting TransRate evaluates both completeness (whether features contain sufficient information of a target task) and compactness (whether features of each class are compact enough for good generalization) of pre-trained features. Theoretically, we have analyzed the close connection of TransRate to the performance after transfer learning. Despite its extraordinary simplicity in 10 lines of codes, TransRate performs remarkably well in extensive evaluations on 35 pre-trained models and 16 downstream tasks.

Meta-learning has emerged as a potent paradigm for quick learning of few-shot tasks, by leveraging the meta-knowledge learned from meta-training tasks. Well-generalized meta-knowledge that facilitates fast adaptation in each task is preferred; however, recent evidence suggests the undesirable memorization effect where the meta-knowledge simply memorizing all meta-training tasks discourages task-specific adaptation and poorly generalizes. There have been several solutions to mitigating the effect, including both regularizer-based and augmentation-based methods, while a systematic understanding of these methods in a single framework is still lacking. In this paper, we offer a novel causal perspective of meta-learning. Through the lens of causality, we conclude the universal label space as a confounder to be the causing factor of memorization and frame the two lines of prevailing methods as different deconfounder approaches. Remarkably, derived from the causal inference principle of front-door adjustment, we propose two frustratingly easy but effective deconfounder algorithms, i.e., sampling multiple versions of the meta-knowledge via Dropout and grouping the meta-knowledge into multiple bins. The proposed causal perspective not only brings in the two deconfounder algorithms that surpass previous works in four benchmark datasets towards combating memorization, but also opens a promising direction for meta-learning.

Recent efforts on text erasing have shown promising results. However, existing methods require rich yet costly label annotations to obtain robust models, which limits their use for practical applications. To this end, we study an unsupervised scenario by proposing a novel Self-supervised Text Erasing (STE) framework that jointly learns to synthesize training images with erasure ground-truth and accurately erase texts in the real world. We first design a style-aware image synthesis function to generate synthetic images with diverse styled texts based on two synthetic mechanisms. To bridge the text style gap between the synthetic and real-world data, a policy network is constructed to control the synthetic mechanisms by picking style parameters with the guidance of two specifically designed rewards. The synthetic training images with ground-truth are then fed to train a coarse-to-fine erasing network. To produce better erasing outputs, a triplet erasure loss is designed to enforce the refinement stage to recover background textures. Moreover, we provide a new dataset (called PosterErase), which contains 60K high-resolution posters and is more challenging for the erasing task. The proposed method has been extensively evaluated with both PosterErase and the widely-used SCUT-Enstext dataset. Notably, on PosterErase, our method achieves 5.07 in terms of FID, with a relative improvement of 20.9% over existing supervised baselines.

Recent years have witnessed the rapid development of meta-learning in improving the meta generalization over tasks in few-shot learning. However, the task-specific level generalization is overlooked in most algorithms. For a novel few-shot learning task where the empirical distribution likely deviates from the true distribution, the model obtained via minimizing the empirical loss can hardly generalize to unseen data. A viable solution to improving the generalization comes as a more accurate approximation of the true distribution; that is, admitting a Gaussian-like vicinal distribution for each of the limited training samples. Thereupon we derive the resulting vicinal loss function over vicinities of all training samples and minimize it instead of the conventional empirical loss over training samples only, favorably free from the exhaustive sampling of all vicinal samples.It remains challenging to obtain the statistical parameters of the vicinal distribution for each sample. To tackle this challenge, we further propose to estimate the statistical parameters as the weighted mean and variance of a set of unlabeled data it passed by a random walk starting from training samples. To verify the performance of the proposed method, we conduct experiments on four standard few-shot learning benchmarks and consolidate the superiority of the proposed method over state-of-the-art few-shot learning baselines.

The success of meta-learning on existing benchmarks is predicated on the assumption that the distribution of meta-training tasks covers meta-testing tasks. Frequent violation of the assumption in applications with either insufficient tasks or a very narrow meta-training task distribution leads to memorization or learner overfitting. Recent solutions have pursued augmentation of meta-training tasks, while it is still an open question to generate both correct and sufficiently imaginary tasks. In this paper, we seek an approach that up-samples meta-training tasks from the task representation via a task up-sampling network. Besides, the resulting approach named Adversarial Task Up-sampling (ATU) suffices to generate tasks that can maximally contribute to the latest meta-learner by maximizing an adversarial loss. On few-shot sine regression and image classification datasets, we empirically validate the marked improvement of ATU over state-of-the-art task augmentation strategies in the meta-testing performance and also the quality of up-sampled tasks.

Retrosynthetic planning occupies a crucial position in synthetic chemistry and, accordingly, drug discovery, which aims to find synthetic pathways of a target molecule through a sequential decision-making process on a set of feasible reactions. While the majority of recent works focus on the prediction of feasible reactions at each step, there have been limited attempts toward improving the sequential decision-making policy. Existing strategies rely on either the expensive and high-variance value estimation by online rollout, or a settled value estimation neural network pre-trained with simulated pathways of limited diversity and no negative feedback. Besides, how to return multiple candidate pathways that are not only diverse but also desirable for chemists (e.g., affordable building block materials) remains an open challenge. To this end, we propose a Goal-dRiven Actor-critic retroSynthetic Planning (GRASP) framework, where we identify the policy that performs goal-driven retrosynthesis navigation toward a user-demand objective. Our experiments on the benchmark Pistachio dataset and a chemists-designed dataset demonstrate that the framework outperforms state-of-the-art approaches by up to 32.2% on search efficiency and 5.6% on quality. Remarkably, our user studies show that GRASP successfully plans pathways that accomplish the goal prescribed with a designated goal (building block materials).

Hyperparathyroidism (HPT) is a symptom of hypertrophy and enlargement of the parathyroids due to the overproduction of parathyroid hormones. Since there are multiple ectopic parathyroid glands, namely the position of the glands is not fixed. The clinical diagnosis and localization of HPT are difficult, which can easily results in underdiagnoses and misdiagnoses. This paper collected and produced two versions of datasets for the computer-aided diagnosis (CAD) of HPT in PASCAL VOC and COCO formats, respectively. Statistical analysis shows that 84.5% of the HPT lesions in the ultrasonic images are medium-scale and large-scale targets. We proposed a hybrid detection network that employs CSPDarknet, spatial pyramid pooling net (SPPnet) and deformable transformer for automatically detecting HPT. We further introduced several attention networks combined with the CNN backbone to improve the network structure. Extensive experiments have been carried out on the datasets by combining various attention networks and the hybrid detection network. The experimental results illustrate that the inclusion of the channel attention mechanisms can improve the performance of HPT lesion detection more effectively than other hybrid networks, especially for large lesion objects. It implies that the channel-wise features are more critical for HPT detection than the global features and the spatial location features.

Sequence labeling aims to predict a finegrained sequence of labels for the text.However, such formulation hinders the effectiveness of supervised methods due to the lack of token-level annotated data.This is exacerbated when we meet a diverse range of languages.In this work, we explore multilingual sequence labeling with minimal supervision using a single unified model for multiple languages.Specifically, we propose a Meta Teacher-Student (MetaTS) Network, a novel meta learning method to alleviate data scarcity by leveraging large multilingual unlabeled data.Prior teacher-student frameworks of self-training rely on rigid teaching strategies, which may hardly produce high-quality pseudo-labels for consecutive and interdependent tokens.On the contrary, MetaTS allows the teacher to dynamically adapt its pseudoannotation strategies by the student's feedback on the generated pseudo-labeled data of each language and thus mitigate error propagation from noisy pseudo-labels.Extensive experiments on both public and real-world multilingual sequence labeling datasets empirically demonstrate the effectiveness of MetaTS 1 .

The surrogate that predicts the performance of hyperparameters has been a key component for sequential model-based hyperparameter optimization. In practical applications, a trial of a hyper-parameter configuration may be so costly that a surrogate is expected to return an optimal configuration with as few trials as possible. Observing that human experts draw on their expertise in a machine learning model by trying configurations that once performed well on other datasets, we are inspired to build a trial-efficient surrogate by transferring the meta-knowledge learned from historical trials on other datasets. We propose an end-to-end surrogate named as Transfer NeuralProcesses (TNP) that learns a comprehensive set of meta-knowledge, including the parameters of historical surrogates, historical trials, and initial configurations for other datasets. Experiments on extensive OpenML datasets and three computer vision datasets demonstrate that the proposed algorithm achieves state-of-the-art performance in at least one order of magnitude less trials.

Meta-learning has proven to be a powerful paradigm for transferring the knowledge from previous tasks to facilitate the learning of a novel task. Current dominant algorithms train a well-generalized model initialization which is adapted to each task via the support set. The crux lies in optimizing the generalization capability of the initialization, which is measured by the performance of the adapted model on the query set of each task. Unfortunately, this generalization measure, evidenced by empirical results, pushes the initialization to overfit the meta-training tasks, which significantly impairs the generalization and adaptation to novel tasks. To address this issue, we actively augment a meta-training task with “more data” when evaluating the generalization. Concretely, we propose two task augmentation methods, including MetaMix and Channel Shuffle. MetaMix linearly combines features and labels of samples from both the support and query sets. For each class of samples, Channel Shuffle randomly replaces a subset of their channels with the corresponding ones from a different class. Theoretical studies show how task augmentation improves the generalization of meta-learning. Moreover, both MetaMix and Channel Shuffle outperform state-of-the-art results by a large margin across many datasets and are compatible with existing meta-learning algorithms.

More recently, there has been a surge of interest in employing machine learning approaches to expedite the drug discovery process where virtual screening for hit discovery and ADMET prediction for lead optimization play essential roles. One of the main obstacles to the wide success of machine learning approaches in these two tasks is that the number of compounds labeled with activities or ADMET properties is too small to build an effective predictive model. This paper seeks to remedy the problem by transferring the knowledge from previous assays, namely in-vivo experiments, by different laboratories and against various target proteins. To accommodate these wildly different assays and capture the similarity between assays, we propose a functional rationalized meta-learning algorithm FRML for such knowledge transfer. FRML constructs the predictive model with layers of neural sub-networks or so-called functional regions. Building on this, FRML shares an initialization for the weights of the predictive model across all assays, while customizes it to each assay with a region localization network choosing the pertinent regions. The compositionality of the model improves the capacity of generalization to various and even out-of-distribution tasks. Empirical results on both virtual screening and ADMET prediction validate the superiority of FRML over state-of-the-art baselines powered with interpretability in assay relationship.

To benefit the learning of a new task, meta-learning has been proposed to transfer a well-generalized meta-model learned from various meta-training tasks. Existing meta-learning algorithms randomly sample meta-training tasks with a uniform probability, under the assumption that tasks are of equal importance. However, it is likely that tasks are detrimental with noise or imbalanced given a limited number of meta-training tasks. To prevent the meta-model from being corrupted by such detrimental tasks or dominated by tasks in the majority, in this paper, we propose an adaptive task scheduler (ATS) for the meta-training process. In ATS, for the first time, we design a neural scheduler to decide which meta-training tasks to use next by predicting the probability being sampled for each candidate task, and train the scheduler to optimize the generalization capacity of the meta-model to unseen tasks. We identify two meta-model-related factors as the input of the neural scheduler, which characterize the difficulty of a candidate task to the meta-model. Theoretically, we show that a scheduler taking the two factors into account improves the meta-training loss and also the optimization landscape. Under the setting of meta-learning with noise and limited budgets, ATS improves the performance on both miniImageNet and a real-world drug discovery benchmark by up to 13% and 18%, respectively, compared to state-of-the-art task schedulers.

Towards the challenging problem of semi-supervised node classification, there have been extensive studies. As a frontier, Graph Neural Networks (GNNs) have aroused great interest recently, which update the representation of each node by aggregating information of its neighbors. However, most GNNs have shallow layers with a limited receptive field and may not achieve satisfactory performance especially when the number of labeled nodes is quite small. To address this challenge, we innovatively propose a graph few-shot learning (GFL) algorithm that incorporates prior knowledge learned from auxiliary graphs to improve classification accuracy on the target graph. Specifically, a transferable metric space characterized by a node embedding and a graph-specific prototype embedding function is shared between auxiliary graphs and the target, facilitating the transfer of structural knowledge. Extensive experiments and ablation studies on four real-world graph datasets demonstrate the effectiveness of our proposed model and the contribution of each component.

Hyperparathyroidism (HPT) is a disorder in which the parathyroid glands produce too much parathyroid hormone (PTH), which may lead to hypocalcemic convulsions, cardiomyopathy, hypertension and other diseases, even threaten the lives of patients under certain severe conditions. Since HPT is usually multiple and ectopic with variable symptoms, the diagnosis and location of HPT is a difficult task even for senior radiologists. A transfer learning-based computer-aided diagnosis (CAD) approach is proposed for automated recognition of HPT in this paper. A dataset of the brightness-mode ultrasound images is developed for the HPT recognition, which is usually annotated by senior radiologists. We addressed the HPT recognition using the various computer vision algorithms on the HPT dataset and obtained good performances for all the algorithms. The experimental results demonstrated that the dataset is effective in aiding the diagnosis of HPT.

The recent emergence of multilingual pretraining language model (mPLM) has enabled breakthroughs on various downstream cross-lingual transfer (CLT) tasks. However, mPLM-based methods usually involve two problems: (1) simply fine-tuning may not adapt general-purpose multilingual representations to be task-aware on low-resource languages; (2) ignore how cross-lingual adaptation happens for downstream tasks. To address the issues, we propose a meta graph learning (MGL) method. Unlike prior works that transfer from scratch, MGL can learn to cross-lingual transfer by extracting meta-knowledge from historical CLT experiences (tasks), making mPLM insensitive to low-resource languages. Besides, for each CLT task, MGL formulates its transfer process as information propagation over a dynamic graph, where the geometric structure can automatically capture intrinsic language relationships to guide cross-lingual transfer explicitly. Empirically, extensive experiments on both public and real-world datasets demonstrate the effectiveness of the MGL method. © 2020 Association for Computational Linguistics

In transfer learning, what and where to transfer has been widely studied. Nevertheless, the learned transfer strategies are at high risk of over-fitting, especially when only a few annotated instances are available in the target domain. In this paper, we introduce the concept of transfer ensemble learning, a new direction to tackle the over-fitting of transfer strategies. Intuitively, models with different transfer strategies offer various perspectives on what and where to transfer. Therefore a core problem is to search these diversely transferred models for ensemble so as to achieve better generalization. Towards this end, we propose the Transferability Slider (TranSlider) for transfer ensemble learning. By decreasing the transferability, we obtain a spectrum of base models ranging from pure exploitation of the source model to unconstrained exploration for the target domain. Furthermore, the manner of decreasing transferability with parameter sharing guarantees fast optimization at no additional training cost. Finally, we conduct extensive experiments with various analyses, which demonstrate that TranSlider achieves the state-of-the-art on comprehensive benchmark datasets.

How to obtain informative representations of molecules is a crucial prerequisite in AI-driven drug design and discovery. Recent researches abstract molecules as graphs and employ Graph Neural Networks (GNNs) for molecular representation learning. Nevertheless, two issues impede the usage of GNNs in real scenarios: (1) insufficient labeled molecules for supervised training; (2) poor generalization capability to new-synthesized molecules. To address them both, we propose a novel framework, GROVER, which stands for Graph Representation frOm self-superVised mEssage passing tRansformer. With carefully designed self-supervised tasks in node-, edge- and graph-level, GROVER can learn rich structural and semantic information of molecules from enormous unlabelled molecular data. Rather, to encode such complex information, GROVER integrates Message Passing Networks into the Transformer-style architecture to deliver a class of more expressive encoders of molecules. The flexibility of GROVER allows it to be trained efficiently on large-scale molecular dataset without requiring any supervision, thus being immunized to the two issues mentioned above. We pre-train GROVER with 100 million parameters on 10 million unlabelled molecules---the biggest GNN and the largest training dataset in molecular representation learning. We then leverage the pre-trained GROVER for molecular property prediction followed by task-specific fine-tuning, where we observe a huge improvement (more than 6% on average) from current state-of-the-art methods on 11 challenging benchmarks. The insights we gained are that well-designed self-supervision losses and largely-expressive pre-trained models enjoy the significant potential on performance boosting.

Many approaches have been proposed for time series forecasting, in light of its significance in wide applications including business demand prediction. However, the existing methods suffer from two key limitations. Firstly, most point prediction models only predict an exact value of each time step without flexibility, which can hardly capture the stochasticity of data. Even probabilistic prediction using the likelihood estimation suffers these problems in the same way. Besides, most of them use the auto-regressive generative mode, where ground-truth is provided during training and replaced by the network’s own one-step ahead output during inference, causing the error accumulation in inference. Thus they may fail to forecast time series for long time horizon due to the error accumulation. To solve these issues, in this paper, we propose a new time series forecasting model -- Adversarial Sparse Transformer (AST), based on Generated Adversarial Networks (GANs). Specifically, AST adopts a Sparse Transformer as the generator to learn a sparse attention map for time series forecasting, and uses a discriminator to improve the prediction performance from sequence level. Extensive experiments on several real-world datasets show the effectiveness and efficiency of our method.

Deep domain adaptation models learn a neural network in an unlabeled target domain by leveraging the knowledge from a labeled source domain. This can be achieved by learning a domain-invariant feature space. Though the learned representations are separable in the source domain, they usually have a large variance and samples with different class labels tend to overlap in the target domain, which yields suboptimal adaptation performance. To fill the gap, a Fisher loss is proposed to learn discriminative representations which are within-class compact and between-class separable. Experimental results on two benchmark datasets show that the Fisher loss is a general and effective loss for deep domain adaptation. Noticeable improvements are brought when it is used together with widely adopted transfer criteria, including MMD, CORAL and domain adversarial loss. For example, an absolute improvement of 6.67% in terms of the mean accuracy is attained when the Fisher loss is used together with the domain adversarial loss on the Office-Home dataset.

Deep learning based medical image diagnosis has shown great potential in clinical medicine. However, it often suffers two major difficulties in real-world applications: 1) only limited labels are available for model training, due to expensive annotation costs over medical images; 2) labeled images may contain considerable label noise (e.g.,mislabeling labels) due to diagnostic difficulties of diseases. To address these, we seek to exploit rich labeled data from relevant domains to help the learning in the target task via Unsupervised Domain Adaptation (UDA). Unlike most UDA methods that rely on clean labeled data or assume samples are equally transferable, we innovatively propose a Collaborative Unsupervised Domain Adaptation algorithm, which conducts transferability-aware adaptation and conquers label noise in a collaborative way. We theoretically analyze the generalization performance of the proposed method, and also empirically evaluate it on both medical and general images. Promising experimental results demonstrate the superiority and generalization of the proposed method.

Aspect-level sentiment classification (ASC) aims at identifying sentiment polarities towards aspects in a sentence, where the aspect can behave as a general Aspect Category (AC) or a specific Aspect Term (AT). However, due to the especially expensive and labor-intensive labeling, existing public corpora in AT-level are all relatively small. Meanwhile, most of the previous methods rely on complicated structures with given scarce data, which largely limits the efficacy of the neural models. In this paper, we exploit a new direction named coarse-to-fine task transfer, which aims to leverage knowledge learned from a rich-resource source domain of the coarse-grained AC task, which is more easily accessible, to improve the learning in a low-resource target domain of the fine-grained AT task. To resolve both the aspect granularity inconsistency and feature mismatch between domains, we propose a Multi-Granularity Alignment Network (MGAN). In MGAN, a novel Coarse2Fine attention guided by an auxiliary task can help the AC task modeling at the same finegrained level with the AT task. To alleviate the feature false alignment, a contrastive feature alignment method is adopted to align aspect-specific feature representations semantically. In addition, a large-scale multi-domain dataset for the AC task is provided. Empirically, extensive experiments demonstrate the effectiveness of the MGAN.

Zheng Li, Xin Li, Ying Wei, Lidong Bing, Yu Zhang, Qiang Yang. Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). 2019.

In order to learn quickly with few samples, meta-learning utilizes prior knowledge learned from previous tasks. However, a critical challenge in meta-learning is task uncertainty and heterogeneity, which can not be handled via globally sharing knowledge among tasks. In this paper, based on gradient-based meta-learning, we propose a hierarchically structured meta-learning (HSML) algorithm that explicitly tailors the transferable knowledge to different clusters of tasks. Inspired by the way human beings organize knowledge, we resort to a hierarchical task clustering structure to cluster tasks. As a result, the proposed approach not only addresses the challenge via the knowledge customization to different clusters of tasks, but also preserves knowledge generalization among a cluster of similar tasks. To tackle the changing of task relationship, in addition, we extend the hierarchical structure to a continual learning environment. The experimental results show that our approach can achieve state-of-the-art performance in both toy-regression and few-shot image classification problems.

Spatial-temporal prediction is a fundamental problem for constructing smart city, which is useful for tasks such as traffic control, taxi dispatching, and environment policy making. Due to data collection mechanism, it is common to see data collection with unbalanced spatial distributions. For example, some cities may release taxi data for multiple years while others only release a few days of data; some regions may have constant water quality data monitored by sensors whereas some regions only have a small collection of water samples. In this paper, we tackle the problem of spatial-temporal prediction for the cities with only a short period of data collection. We aim to utilize the long-period data from other cities via transfer learning. Different from previous studies that transfer knowledge from one single source city to a target city, we are the first to leverage information from multiple cities to increase the stability of transfer. Specifically, our proposed model is designed as a spatial-temporal network with a meta-learning paradigm. The meta-learning paradigm learns a well-generalized initialization of the spatial-temporal network, which can be effectively adapted to target cities. In addition, a pattern-based spatial-temporal memory is designed to distill long-term temporal information (i.e., periodicity). We conduct extensive experiments on two tasks: traffic (taxi and bike) prediction and water quality prediction. The experiments demonstrate the effectiveness of our proposed model over several competitive baseline models.

Cross-domain sentiment classification aims to leverage useful information in a source domain to help do sentiment classification in a target domain that has no or little supervised information. Existing cross-domain sentiment classification methods cannot automatically capture non-pivots, i.e., the domain-specific sentiment words, and pivots, i.e., the domain-shared sentiment words, simultaneously. In order to solve this problem, we propose a Hierarchical Attention Transfer Network (HATN) for cross-domain sentiment classification. The proposed HATN provides a hierarchical attention transfer mechanism which can transfer attentions for emotions across domains by automatically capturing pivots and non-pivots. Besides, the hierarchy of the attention mechanism mirrors the hierarchical structure of documents, which can help locate the pivots and non-pivots better. The proposed HATN consists of two hierarchical attention networks, with one named P-net aiming to find the pivots and the other named NP-net aligning the non-pivots by using the pivots as a bridge. Specifically, P-net firstly conducts individual attention learning to provide positive and negative pivots for NP-net. Then, P-net and NP-net conduct joint attention learning such that the HATN can simultaneously capture pivots and non-pivots and realize transferring attentions for emotions across domains. Experiments on the Amazon review dataset demonstrate the effectiveness of HATN.

Traditional recommendation systems (RecSys) suffer from two problems: the exploitation-exploration dilemma and the cold-start problem. One solution to solving the exploitation-exploration dilemma is the contextual bandit policy, which adaptively exploits and explores user interests. As a result, the contextual bandit policy achieves increased rewards in the long run. The contextual bandit policy, however, may cause the system to explore more than needed in the cold-start situations, which can lead to worse short-term rewards. Cross-domain RecSys methods adopt transfer learning to leverage prior knowledge in a source RecSys domain to jump start the cold-start target RecSys. To solve the two problems together, in this paper, we propose the first applicable transferable contextual bandit (TCB) policy for the cross-domain recommendation. TCB not only benefits the exploitation but also accelerates the exploration in the target RecSys. TCB's exploration, in turn, helps to learn how to transfer between different domains. TCB is a general algorithm for both homogeneous and heterogeneous domains. We perform both theoretical regret analysis and empirical experiments. The empirical results show that TCB outperforms the state-of-the-art algorithms over time.

In transfer learning, what and how to transfer are two primary issues to be addressed, as different transfer learning algorithms applied between a source and a target domain result in different knowledge transferred and thereby the performance improvement in the target domain. Determining the optimal one that maximizes the performance improvement requires either exhaustive exploration or considerable expertise. Meanwhile, it is widely accepted in educational psychology that human beings improve transfer learning skills of deciding what to transfer through meta-cognitive reflection on inductive transfer learning practices. Motivated by this, we propose a novel transfer learning framework known as Learning to Transfer (L2T) to automatically determine what and how to transfer are the best by leveraging previous transfer learning experiences. We establish the L2T framework in two stages: 1) we learn a reflection function encrypting transfer learning skills from experiences; and 2) we infer what and how to transfer are the best for a future pair of domains by optimizing the reflection function. We also theoretically analyse the algorithmic stability and generalization bound of L2T, and empirically demonstrate its superiority over several state-of-the-art transfer learning algorithms.

Multitask learning has shown promising performance in many applications and many multitask models have been proposed. In order to identify an effective multitask model for a given multitask problem, we propose a learning framework called Learning to MultiTask (L2MT). To achieve the goal, L2MT exploits historical multitask experience which is organized as a training set consisting of several tuples, each of which contains a multitask problem with multiple tasks, a multitask model, and the relative test error. Based on such training set, L2MT first uses a proposed layerwise graph neural network to learn task embeddings for all the tasks in a multitask problem and then learns an estimation function to estimate the relative test error based on task embeddings and the representation of the multitask model based on a unified formulation. Given a new multitask problem, the estimation function is used to identify a suitable multitask model. Experiments on benchmark datasets show the effectiveness of the proposed L2MT framework.

Domain adaptation tasks such as cross-domain sentiment classification have raised much attention in recent years. Due to the domain discrepancy, a sentiment classifier trained in a source domain may not work well when directly applied to a target domain. Traditional methods need to manually select pivots, which behave in the same way for discriminative learning in both domains. Recently, deep learning methods have been proposed to learn a representation shared by domains. However, they lack the interpretability to directly identify the pivots. To address the problem, we introduce an end-to-end Adversarial Memory Network (AMN) for cross-domain sentiment classification. Unlike existing methods, our approach can automatically capture the pivots using an attention mechanism. Our framework consists of two parameter-shared memory networks: one is for sentiment classification and the other is for domain classification. The two networks are jointly trained so that the selected features minimize the sentiment classification error and at the same time make the domain classifier indiscriminative between the representations from the source or target domains. Moreover, unlike deep learning methods that cannot tell us which words are the pivots, our approach can offer a direct visualization of them. Experiments on the Amazon review dataset demonstrate that our approach can significantly outperform state-of-the-art methods.

Deep neural networks (DNN) have achieved breakthroughs in applications with large sample size. However, when facing high dimension, low sample size (HDLSS) data, such as the phenotype prediction problem using genetic data in bioinformatics, DNN suffers from overfitting and high-variance gradients. In this paper, we propose a DNN model tailored for the HDLSS data, named Deep Neural Pursuit (DNP). DNP selects a subset of high dimensional features for the alleviation of overfitting and takes the average over multiple dropouts to calculate gradients with low variance. As the first DNN method applied on the HDLSS data, DNP enjoys the advantages of the high nonlinearity, the robustness to high dimensionality, the capability of learning from a small number of samples, the stability in feature selection, and the end-to-end training. We demonstrate these advantages of DNP via empirical results on both synthetic and real-world biological datasets.

Ubiquitous computing tasks, such as human activity recognition (HAR), are enabling a wide spectrum of applications, ranging from healthcare to environment monitoring. The success of a ubiquitous computing task relies on sufficient physical sensor data with groundtruth labels, which are always scarce due to the expensive annotating process. Meanwhile, social media platforms provide a lot of social or semantic context information. People share what they are doing and where they are frequently in the messages they post. This rich set of socially shared activities motivates us to transfer knowledge from social media to address the sparsity issue of labelled physical sensor data. In order to transfer the knowledge of social and semantic context, we propose a Co-Regularized Heterogeneous Transfer Learning (CoHTL) model, which builds a common semantic space derived from two heterogeneous domains. Our proposed method outperforms state-of-the-art methods on two ubiquitous computing tasks, namely human activity recognition and region function discovery.

The rapid urbanization has motivated extensive research on urban computing. It is critical for urban computing tasks to unlock the power of the diversity of data modalities generated by different sources in urban spaces, such as vehicles and humans. However, we are more likely to encounter the label scarcity problem and the data insufficiency problem when solving an urban computing task in a city where services and infrastructures are not ready or just built. In this paper, we propose a FLexible multimOdal tRAnsfer Learning (FLORAL) method to transfer knowledge from a city where there exist sufficient multimodal data and labels, to this kind of cities to fully alleviate the two problems. FLORAL learns semantically related dictionaries for multiple modalities from a source domain, and simultaneously transfers the dictionaries and labelled instances from the source into a target domain. We evaluate the proposed method with a case study of air quality prediction.

Multi-modal similarity search has attracted considerable attention to meet the need of information retrieval across different types of media. To enable efficient multi-modal similarity search in large-scale databases recently, researchers start to study multi-modal hashing. Most of the existing methods are applied to search across multi-views among which explicit correspondence is provided. Given a multi-modal similarity search task, we observe that abundant multi-view data can be found on the Web which can serve as an auxiliary bridge. In this paper, we propose a Heterogeneous Translated Hashing (HTH) method with such auxiliary bridge incorporated not only to improve current multi-view search but also to enable similarity search across heterogeneous media which have no direct correspondence. HTH provides more flexible and discriminative ability by embedding heterogeneous media into different Hamming spaces, compared to almost all existing methods that map heterogeneous data in a common Hamming space. We formulate a joint optimization model to learn hash functions embedding heterogeneous media into different Hamming spaces, and a translator aligning different Hamming spaces. The extensive experiments on two real-world datasets, one publicly available dataset of Flickr, and the other MIRFLICKR-Yahoo Answers dataset, highlight the effectiveness and efficiency of our algorithm.

Hashing has enjoyed a great success in large-scale similarity search. Recently, researchers have studied the multi-modal hashing to meet the need of similarity search across different types of media. However, most of the existing methods are applied to search across multi-views among which explicit bridge information is provided. Given a heterogeneous media search task, we observe that abundant multi-view data can be found on the Web which can serve as an auxiliary bridge. In this paper, we propose a Heterogeneous Translated Hashing (HTH) method with such auxiliary bridge incorporated not only to improve current multi-view search but also to enable similarity search across heterogeneous media which have no direct correspondence. HTH simultaneously learns hash functions embedding heterogeneous media into different Hamming spaces, and translators aligning these spaces. Unlike almost all existing methods that map heterogeneous data in a common Hamming space, mapping to different spaces provides more flexible and discriminative ability. We empirically verify the effectiveness and efficiency of our algorithm on two real world large datasets, one publicly available dataset of Flickr and the other MIRFLICKR-Yahoo Answers dataset.