Vision-language models (VLMs) like CLIP have demonstrated remarkable applicability across a variety of downstream tasks, including zero-shot image classification. Recently, the use of prompts or adapters for efficient transfer learning (ETL) has gained significant attention for effectively adapting to downstream tasks. However, previous studies have overlooked the challenge of varying transfer difficulty of downstream tasks. In this paper, we empirically analyze how each ETL method behaves with respect to transfer difficulty. Our observations indicate that utilizing vision prompts and text adapters is crucial for adaptability and generalizability in domains with high difficulty. Also, by applying an adaptive ensemble approach that integrates task-adapted VLMs with pre-trained VLMs and strategically leverages more general knowledge in low-difficulty and less in high-difficulty domains, we consistently enhance performance across both types of domains. Based on these observations, we propose an adaptive ensemble method that combines visual prompts and text adapters with pre-trained VLMs, tailored by transfer difficulty, to achieve optimal performance for any target domain. Upon experimenting with extensive benchmarks, our method consistently outperforms all baselines, particularly on unseen tasks, demonstrating its effectiveness.
Accurate Defect detection is crucial for ensuring the trustworthiness of intelligent railway systems. Current approaches rely on single deep-learning models, like CNNs, which employ a large amount of data to capture underlying patterns. Training a new defect classifier with limited samples often leads to overfitting and poor performance on unseen images. To address this, researchers have advocated transfer learning and fine-tuning the pre-trained models. However, using a single backbone network in transfer learning still may cause bottleneck issues and inconsistent performance if it is not suitable for a specific problem domain. To overcome these challenges, we propose a reusable AI-enabled defect detection approach. By combining ensemble learning with transfer learning models (VGG-19, MobileNetV3, and ResNet-50), we improved the classification accuracy and achieved consistent performance at a certain phase of training. Our empirical analysis demonstrates better and more consistent performance compared to other state-of-the-art approaches. The consistency substantiates the reusability of the defect detection system for newly evolved defected rail parts. Therefore we anticipate these findings to benefit further research and development of reusable AI-enabled solutions for railway systems.
Quantum computing has the potential to solve problems that are intractable for classical systems, yet the high error rates in contemporary quantum devices often exceed tolerable limits for useful algorithm execution. Quantum Error Correction (QEC) mitigates this by employing redundancy, distributing quantum information across multiple data qubits and utilizing syndrome qubits to monitor their states for errors. The syndromes are subsequently interpreted by a decoding algorithm to identify and correct errors in the data qubits. This task is complex due to the multiplicity of error sources affecting both data and syndrome qubits as well as syndrome extraction operations. Additionally, identical syndromes can emanate from different error sources, necessitating a decoding algorithm that evaluates syndromes collectively. Although machine learning (ML) decoders such as multi-layer perceptrons (MLPs) and convolutional neural networks (CNNs) have been proposed, they often focus on local syndrome regions and require retraining when adjusting for different code distances. We introduce a transformer-based QEC decoder which employs self-attention to achieve a global receptive field across all input syndromes. It incorporates a mixed loss training approach, combining both local physical error and global parity label losses. Moreover, the transformer architecture's inherent adaptability to variable-length inputs allows for efficient transfer learning, enabling the decoder to adapt to varying code distances without retraining. Evaluation on six code distances and ten different error configurations demonstrates that our model consistently outperforms non-ML decoders, such as Union Find (UF) and Minimum Weight Perfect Matching (MWPM), and other ML decoders, thereby achieving best logical error rates. Moreover, the transfer learning can save over 10x of training cost.
Data-driven models achieve successful results in Speech Emotion Recognition (SER). However, these models, which are often based on general acoustic features or end-to-end approaches, show poor performance when the testing set has a different language than the training set or when these sets are taken from different datasets. To alleviate these problems, this paper presents an end-to-end Deep Neural Network (DNN) model based on transfer learning for cross-language and cross-corpus SER. We use the wav2vec 2.0 pre-trained model to transform audio time-domain waveforms from different languages, different speakers and different recording conditions into a feature space shared by multiple languages, thereby reducing the language variabilities in the speech embeddings. Next, we propose a new Deep-Within-Class Covariance Normalisation (Deep-WCCN) layer that can be inserted into the DNN model and aims to reduce other variabilities including speaker variability, channel variability and so on. The entire model is fine-tuned in an end-to-end manner on a combined loss and is validated on datasets from three languages (i.e. English, German, Chinese). Experimental results show that our proposed method outperforms the baseline model that is based on common acoustic feature sets for SER in the within-language setting and the cross-language setting. In addition, we also experimentally validate the effectiveness of Deep-WCCN, which can further improve the model performance. Next, we show that the proposed transfer learning method has good data efficiency when merging target language data into the fine-tuning process. The model speaker-independent SER performance increases with up to 15.6% when only 160s of target language data is used. Finally, our proposed model shows significantly better performance than other state-of-the-art models in cross-language SER.
Transfer learning for Bayesian optimisation has generally assumed a strong similarity between optimisation tasks, with at least a subset having similar optimal inputs. This assumption can reduce computational costs, but it is violated in a wide range of optimisation problems where transfer learning may nonetheless be useful. We replace this assumption with a weaker one only requiring the shape of the optimisation landscape to be similar, and analyse the recent method Prior Learning for Bayesian Optimisation - PLeBO - in this setting. By learning priors for the hyperparameters of the Gaussian process surrogate model we can better approximate the underlying function, especially for few function evaluations. We validate the learned priors and compare to a breadth of transfer learning approaches, using synthetic data and a recent air pollution optimisation problem as benchmarks. We show that PLeBO and prior transfer find good inputs in fewer evaluations.
Machine translation (MT) for low-resource languages such as Ge'ez, an ancient language that is no longer the native language of any community, faces challenges such as out-of-vocabulary words, domain mismatches, and lack of sufficient labeled training data. In this work, we explore various methods to improve Ge'ez MT, including transfer-learning from related languages, optimizing shared vocabulary and token segmentation approaches, finetuning large pre-trained models, and using large language models (LLMs) for few-shot translation with fuzzy matches. We develop a multilingual neural machine translation (MNMT) model based on languages relatedness, which brings an average performance improvement of about 4 BLEU compared to standard bilingual models. We also attempt to finetune the NLLB-200 model, one of the most advanced translation models available today, but find that it performs poorly with only 4k training samples for Ge'ez. Furthermore, we experiment with using GPT-3.5, a state-of-the-art LLM, for few-shot translation with fuzzy matches, which leverages embedding similarity-based retrieval to find context examples from a parallel corpus. We observe that GPT-3.5 achieves a remarkable BLEU score of 9.2 with no initial knowledge of Ge'ez, but still lower than the MNMT baseline of 15.2. Our work provides insights into the potential and limitations of different approaches for low-resource and ancient language MT.
Previous efforts on hyperparameter optimization (HPO) of machine learning (ML) models predominately focus on algorithmic advances, yet little is known about the topography of the underlying hyperparameter (HP) loss landscape, which plays a fundamental role in governing the search process of HPO. While several works have conducted fitness landscape analysis (FLA) on various ML systems, they are limited to properties of isolated landscape without interrogating the potential structural similarities among them. The exploration of such similarities can provide a novel perspective for understanding the mechanism behind modern HPO methods, but has been missing, possibly due to the expensive cost of large-scale landscape construction, and the lack of effective analysis methods. In this paper, we mapped 1,500 HP loss landscapes of 6 representative ML models on 63 datasets across different fidelity levels, with 11M+ configurations. By conducting exploratory analysis on these landscapes with fine-grained visualizations and dedicated FLA metrics, we observed a similar landscape topography across a wide range of models, datasets, and fidelities, and shed light on several central topics in HPO.
Deep learning (DL) has shown great potential in revolutionizing the traditional communications system. Many applications in communications have adopted DL techniques due to their powerful representation ability. However, the learning-based methods can be dependent on the training dataset and perform worse on unseen interference due to limited model generalizability and complexity. In this paper, we consider the semantic communication (SemCom) system with multiple users, where there is a limited number of training samples and unexpected interference. To improve the model generalization ability and reduce the model size, we propose a knowledge distillation (KD) based system where Transformer based encoder-decoder is implemented as the semantic encoder-decoder and fully connected neural networks are implemented as the channel encoder-decoder. Specifically, four types of knowledge transfer and model compression are analyzed. Important system and model parameters are considered, including the level of noise and interference, the number of interfering users and the size of the encoder and decoder. Numerical results demonstrate that KD significantly improves the robustness and the generalization ability when applied to unexpected interference, and it reduces the performance loss when compressing the model size.
Knowledge distillation (KD) enables a smaller "student" model to mimic a larger "teacher" model by transferring knowledge from the teacher's output or features. However, most KD methods treat all samples uniformly, overlooking the varying learning value of each sample and thereby limiting effectiveness. In this paper, we propose Entropy-based Adaptive Knowledge Distillation (EA-KD), a simple yet effective plug-and-play KD method that prioritizes learning from valuable samples. EA-KD quantifies each sample's learning value by strategically combining the entropy of the teacher and student output, then dynamically reweights the distillation loss to place greater emphasis on high-value samples. Extensive experiments across diverse KD frameworks and tasks$\unicode{x2014}$including image classification, object detection, and large language model (LLM) distillation$\unicode{x2014}$demonstrate that EA-KD consistently enhances performance, achieving state-of-the-art results with negligible computational cost. Our code will be publicly available.
Zero-shot 3D part segmentation is a challenging and fundamental task. In this work, we propose a novel pipeline, ZeroPS, which achieves high-quality knowledge transfer from 2D pretrained foundation models (FMs), SAM and GLIP, to 3D object point clouds. We aim to explore the natural relationship between multi-view correspondence and the FMs' prompt mechanism and build bridges on it. In ZeroPS, the relationship manifests as follows: 1) lifting 2D to 3D by leveraging co-viewed regions and SAM's prompt mechanism, 2) relating 1D classes to 3D parts by leveraging 2D-3D view projection and GLIP's prompt mechanism, and 3) enhancing prediction performance by leveraging multi-view observations. Extensive evaluations on the PartNetE and AKBSeg benchmarks demonstrate that ZeroPS significantly outperforms the SOTA method across zero-shot unlabeled and instance segmentation tasks. ZeroPS does not require additional training or fine-tuning for the FMs. ZeroPS applies to both simulated and real-world data. It is hardly affected by domain shift. The project page is available at https://luis2088.github.io/ZeroPS_page.
Accounting for the increased concern for public safety, automatic abnormal event detection and recognition in a surveillance scene is crucial. It is a current open study subject because of its intricacy and utility. The identification of aberrant events automatically, it's a difficult undertaking because everyone's idea of abnormality is different. A typical occurrence in one circumstance could be seen as aberrant in another. Automatic anomaly identification becomes particularly challenging in the surveillance footage with a large crowd due to congestion and high occlusion. With the use of machine learning techniques, this thesis study aims to offer the solution for this use case so that human resources won't be required to keep an eye out for any unusual activity in the surveillance system records. We have developed a novel generative adversarial network (GAN) based anomaly detection model. This model is trained such that it learns together about constructing a high dimensional picture space and determining the latent space from the video's context. The generator uses a residual Autoencoder architecture made up of a multi-stage channel attention-based decoder and a two-stream, deep convolutional encoder that can realise both spatial and temporal data. We have also offered a technique for refining the GAN model that reduces training time while also generalising the model by utilising transfer learning between datasets. Using a variety of assessment measures, we compare our model to the current state-of-the-art techniques on four benchmark datasets. The empirical findings indicate that, in comparison to existing techniques, our network performs favourably on all datasets.
Quantum neural networks (QNNs), harnessing superposition and entanglement, have shown potential to surpass classical methods in complex learning tasks but remain limited by hardware constraints and noisy conditions. In this work, we present a novel framework for transferring knowledge from classical convolutional neural networks (CNNs) to QNNs via knowledge distillation, thereby reducing the need for resource intensive quantum training and error mitigation. We conduct extensive experiments using two parameterized quantum circuits (PQCs) with 4 and 8 qubits on MNIST, Fashion MNIST, and CIFAR10 datasets. The approach demonstrates consistent accuracy improvements attributed to distilled knowledge from larger classical networks. Through ablation studies, we systematically compare the effect of state of the art dimensionality reduction techniques fully connected layers, center cropping, principal component analysis, and pooling to compress high-dimensional image data prior to quantum encoding. Our findings reveal that fully connected layers retain the most salient features for QNN inference, thereby surpassing other down sampling approaches. Additionally, we examine state of the art data encoding methods (amplitude, angle, and qubit encoding) and identify amplitude encoding as the optimal strategy, yielding superior accuracy across all tested datasets and qubit configurations. Through computational analyses, we show that our distilled 4-qubit and 8-qubit QNNs achieve competitive performance while utilizing significantly fewer parameters than their classical counterparts. Our results establish a promising paradigm for bridging classical deep learning and emerging quantum computing, paving the way for more powerful, resource conscious models in quantum machine intelligence.
A growing number of software projects makes it increasingly crucial to predict software defects. If adequate historical data are accessible, within-project defect prediction models can be reliable. However, during the early phases of software development, sufficient data are not available to train an effective predictor. Cross-project defect prediction (CPDP) utilizes information from previous mature projects (source data) for predicting whether new software modules (target data) will be defective. CPDP models must take into account the fact that data distributions between source and target projects are different. Cross-project defect prediction often reduces distribution differences by either selecting training data or using transfer learning methods. Using transfer learning effectively reduces distribution differences in recent CPDP models, yet none of them have taken into account the possibility that negative transfer may occur as a result of the imbalanced nature of defect data. In this paper, a four-step model is proposed, of which three steps are dedicated to the preparation of training data and their initial weights for use in the fourth step, which involves an enhanced version of the transfer boosting algorithm. In this algorithm, the imbalance nature of data is considered and the weighting of the source data is updated to enhance the prediction performance. Therefore, aside from reducing distribution discrepancy between source and target data, this model also addresses the issues related to defect data class imbalance. As compared to four state-of-the-art CPDP models, this model provided consistent and accurate predictions for fifteen projects from PROMISE, AEEEM, and SOFTLAB. Our proposed model provided the best average results for both AUC and F-measure and in some datasets, the improvements were more than 5%.
Traditional surveillance systems rely on human attention, limiting their effectiveness. This study employs convolutional neural networks and transfer learning to develop a real-time computer vision system for automatic handgun detection. Comprehensive analysis of online handgun detection methods is conducted, emphasizing reducing false positives and learning time. Transfer learning is demonstrated as an effective approach. Despite technical challenges, the proposed system achieves a precision rate of 84.74%, demonstrating promising performance comparable to related works, enabling faster learning and accurate automatic handgun detection for enhanced security. This research advances security measures by reducing human monitoring dependence, showcasing the potential of transfer learning-based approaches for efficient and reliable handgun detection.
Unsupervised Domain Adaptive Semantic Segmentation (UDA-SS) aims to transfer the supervision from a labeled source domain to an unlabeled target domain. The majority of existing UDA-SS works typically consider images whilst recent attempts have extended further to tackle videos by modeling the temporal dimension. Although the two lines of research share the major challenges -- overcoming the underlying domain distribution shift, their studies are largely independent, resulting in fragmented insights, a lack of holistic understanding, and missed opportunities for cross-pollination of ideas. This fragmentation prevents the unification of methods, leading to redundant efforts and suboptimal knowledge transfer across image and video domains. Under this observation, we advocate unifying the study of UDA-SS across video and image scenarios, enabling a more comprehensive understanding, synergistic advancements, and efficient knowledge sharing. To that end, we explore the unified UDA-SS from a general data augmentation perspective, serving as a unifying conceptual framework, enabling improved generalization, and potential for cross-pollination of ideas, ultimately contributing to the overall progress and practical impact of this field of research. Specifically, we propose a Quad-directional Mixup (QuadMix) method, characterized by tackling distinct point attributes and feature inconsistencies through four-directional paths for intra- and inter-domain mixing in a feature space. To deal with temporal shifts with videos, we incorporate optical flow-guided feature aggregation across spatial and temporal dimensions for fine-grained domain alignment. Extensive experiments show that our method outperforms the state-of-the-art works by large margins on four challenging UDA-SS benchmarks. Our source code and models will be released at https://github.com/ZHE-SAPI/UDASS.
As a surrogate for computationally intensive meso-scale simulation of woven composites, this article presents Recurrent Neural Network (RNN) models. Leveraging the power of transfer learning, the initialization challenges and sparse data issues inherent in cyclic shear strain loads are addressed in the RNN models. A mean-field model generates a comprehensive data set representing elasto-plastic behavior. In simulations, arbitrary six-dimensional strain histories are used to predict stresses under random walking as the source task and cyclic loading conditions as the target task. Incorporating sub-scale properties enhances RNN versatility. In order to achieve accurate predictions, the model uses a grid search method to tune network architecture and hyper-parameter configurations. The results of this study demonstrate that transfer learning can be used to effectively adapt the RNN to varying strain conditions, which establishes its potential as a useful tool for modeling path-dependent responses in woven composites.
In collaborative human-robot manipulation, a robot must predict human intents and adapt its actions accordingly to smoothly execute tasks. However, the human's intent in turn depends on actions the robot takes, creating a chicken-or-egg problem. Prior methods ignore such inter-dependency and instead train marginal intent prediction models independent of robot actions. This is because training conditional models is hard given a lack of paired human-robot interaction datasets. Can we instead leverage large-scale human-human interaction data that is more easily accessible? Our key insight is to exploit a correspondence between human and robot actions that enables transfer learning from human-human to human-robot data. We propose a novel architecture, InteRACT, that pre-trains a conditional intent prediction model on large human-human datasets and fine-tunes on a small human-robot dataset. We evaluate on a set of real-world collaborative human-robot manipulation tasks and show that our conditional model improves over various marginal baselines. We also introduce new techniques to tele-operate a 7-DoF robot arm and collect a diverse range of human-robot collaborative manipulation data, which we open-source.
Active Domain Adaptation (ADA) aims to maximally boost model adaptation in a new target domain by actively selecting a limited number of target data to annotate.This setting neglects the more practical scenario where training data are collected from multiple sources. This motivates us to target a new and challenging setting of knowledge transfer that extends ADA from a single source domain to multiple source domains, termed Multi-source Active Domain Adaptation (MADA). Not surprisingly, we find that most traditional ADA methods cannot work directly in such a setting, mainly due to the excessive domain gap introduced by all the source domains and thus their uncertainty-aware sample selection can easily become miscalibrated under the multi-domain shifts. Considering this, we propose a Dynamic integrated uncertainty valuation framework(Detective) that comprehensively consider the domain shift between multi-source domains and target domain to detect the informative target samples. Specifically, the leverages a dynamic Domain Adaptation(DA) model that learns how to adapt the model's parameters to fit the union of multi-source domains. This enables an approximate single-source domain modeling by the dynamic model. We then comprehensively measure both domain uncertainty and predictive uncertainty in the target domain to detect informative target samples using evidential deep learning, thereby mitigating uncertainty miscalibration. Furthermore, we introduce a contextual diversity-aware calculator to enhance the diversity of the selected samples. Experiments demonstrate that our solution outperforms existing methods by a considerable margin on three domain adaptation benchmarks.
Given the increasing availability of RNA-seq data and its complex and heterogeneous nature, there has been growing interest in applying AI/machine learning methodologies to work with such data modalities. However, because omics data is characterized by high dimensionality and low sample size (HDLSS), current attempts at integrating AI in this domain require significant human guidance and expertise to mitigate overfitting. In this work we look at how transfer learning can be improved to learn from small RNA-seq sample sizes without significant human interference. The strategy is to gain general prior knowledge about a particular domain of data (e.g. RNA-seq data) by pre-training on a general task with a large aggregate of data, then fine-tuning to various specific, downstream target tasks in the same domain. Because previous attempts have shown traditional transfer learning failing on HLDSS, we propose to improve performance by using Guided Transfer Learning (GTL). Collaborating with Robots Go Mental, the AI we deploy here not only learns good initial parameters during pre-training, but also learns inductive biases that affect how the AI learns downstream tasks. In this approach, we first pre-trained on recount3 data, a collection of over 400,000 mouse RNA-seq samples sourced from thousands of individual studies. With such a large collection, patterns of expression between the ~30,000 genes in mammalian systems were pre-determined. Such patterns were sufficient for the pre-trained AI agent to efficiently learn new downstream tasks involving RNA-seq datasets with very low sample sizes and performed notably better on few-shot learning tasks compared to the same model without pre-training.
Accurate and robust drug response prediction is of utmost importance in precision medicine. Although many models have been developed to utilize the representations of drugs and cancer cell lines for predicting cancer drug responses (CDR), their performances can be improved by addressing issues such as insufficient data modality, suboptimal fusion algorithms, and poor generalizability for novel drugs or cell lines. We introduce TransCDR, which uses transfer learning to learn drug representations and fuses multi-modality features of drugs and cell lines by a self-attention mechanism, to predict the IC50 values or sensitive states of drugs on cell lines. We are the first to systematically evaluate the generalization of the CDR prediction model to novel (i.e., never-before-seen) compound scaffolds and cell line clusters. TransCDR shows better generalizability than 8 state-of-the-art models. TransCDR outperforms its 5 variants that train drug encoders (i.e., RNN and AttentiveFP) from scratch under various scenarios. The most critical contributors among multiple drug notations and omics profiles are Extended Connectivity Fingerprint and genetic mutation. Additionally, the attention-based fusion module further enhances the predictive performance of TransCDR. TransCDR, trained on the GDSC dataset, demonstrates strong predictive performance on the external testing set CCLE. It is also utilized to predict missing CDRs on GDSC. Moreover, we investigate the biological mechanisms underlying drug response by classifying 7,675 patients from TCGA into drug-sensitive or drug-resistant groups, followed by a Gene Set Enrichment Analysis. TransCDR emerges as a potent tool with significant potential in drug response prediction. The source code and data can be accessed at https://github.com/XiaoqiongXia/TransCDR.