Efficient Adapter Finetuning for Tail Languages in Streaming Multilingual ASR

The end-to-end ASR model is often desired in the streaming multilingual scenario since it is easier to deploy and can benefit from pre-trained speech models such as powerful foundation models. Meanwhile, the heterogeneous nature and imbalanced data abundance of different languages may cause performance degradation, leading to asynchronous peak performance for different languages during training, especially on tail ones. Sometimes even the data itself may become unavailable as a result of the enhanced privacy protection. Existing work tend to significantly increase the model size or learn language-specific decoders to accommodate each language separately. In this study, we explore simple yet effective Language-Dependent Adapter (LDA) finetuning under a cascaded Conformer transducer framework enhanced by teacher pseudo-labeling for tail languages in the streaming multilingual ASR. The adapter only accounts for 0.4% of the full model per language. It is plugged into the frozen foundation model and is the only trainable module during the finetuning process with noisy student training. The final model merges the adapter parameters from different checkpoints for different languages. The model performance is validated on a challenging multilingual dictation dataset, which includes 39 tail languages across Latin, Greek, Arabic, etc. Our proposed method brings 12.2% word error rate reduction on average and up to 37.5% on a single locale. Furthermore, we show that our parameter-efficient LDA can match the quality of the full model finetuning, thus greatly alleviating the asynchronous peak performance issue.

VOT: Revolutionizing Speaker Verification with Memory and Attention Mechanisms

Speaker verification is to judge the similarity of two unknown voices in an open set, where the ideal speaker embedding should be able to condense discriminant information into a compact utterance-level representation that has small intra-speaker distances and large inter-speaker distances.We propose a novel model named Voice Transformer(VOT) for speaker verification. The model consists of multiple parallel Transformers, and the outputs of these Transformers are adaptively combined. Deeply-Fused Semantic Memory Network(DFSMN)is integrated into the attention parts of these Transformers to capture long-distance information and enhance the local dependencies. Statistical pooling layers are incorporated to enhance overall performance without significantly increasing the number of parameters. We propose a new loss function called Additive Angular Margin Focal Loss(AAMF) to address the hard sample mining issue.We evaluate the proposed approach on the VoxCeleb1 and CN-Celeb2 datasets. The experimental results demonstrate that VOT achieves state-of-the-art results, outperforming nearly all existing models. The code is available on GitHub.

Crafter: Facial Feature Crafting against Inversion-based Identity Theft on Deep Models

With the increased capabilities at the edge (e.g., mobile device) and more stringent privacy requirement, it becomes a recent trend for deep learning-enabled applications to pre-process sensitive raw data at the edge and transmit the features to the backend cloud for further processing. A typical application is to run machine learning (ML) services on facial images collected from different individuals. To prevent identity theft, conventional methods commonly rely on an adversarial game-based approach to shed the identity information from the feature. However, such methods can not defend against adaptive attacks, in which an attacker takes a countermove against a known defence strategy. We propose Crafter, a feature crafting mechanism deployed at the edge, to protect the identity information from adaptive model inversion attacks while ensuring the ML tasks are properly carried out in the cloud. The key defence strategy is to mislead the attacker to a non-private prior from which the attacker gains little about the private identity. In this case, the crafted features act like poison training samples for attackers with adaptive model updates. Experimental results indicate that Crafter successfully defends both basic and possible adaptive attacks, which can not be achieved by state-of-the-art adversarial game-based methods.

Convolutional Neural Network Ensemble Learning for Hyperspectral Imaging-based Blackberry Fruit Ripeness Detection in Uncontrolled Farm Environment

Fruit ripeness estimation models have for decades depended on spectral index features or colour-based features, such as mean, standard deviation, skewness, colour moments, and/or histograms for learning traits of fruit ripeness. Recently, few studies have explored the use of deep learning techniques to extract features from images of fruits with visible ripeness cues. However, the blackberry (Rubus fruticosus) fruit does not show obvious and reliable visible traits of ripeness when mature and therefore poses great difficulty to fruit pickers. The mature blackberry, to the human eye, is black before, during, and post-ripening. To address this engineering application challenge, this paper proposes a novel multi-input convolutional neural network (CNN) ensemble classifier for detecting subtle traits of ripeness in blackberry fruits. The multi-input CNN was created from a pre-trained visual geometry group 16-layer deep convolutional network (VGG16) model trained on the ImageNet dataset. The fully connected layers were optimized for learning traits of ripeness of mature blackberry fruits. The resulting model served as the base for building homogeneous ensemble learners that were ensemble using the stack generalization ensemble (SGE) framework. The input to the network is images acquired with a stereo sensor using visible and near-infrared (VIS-NIR) spectral filters at wavelengths of 700 nm and 770 nm. Through experiments, the proposed model achieved 95.1% accuracy on unseen sets and 90.2% accuracy with in-field conditions. Further experiments reveal that machine sensory is highly and positively correlated to human sensory over blackberry fruit skin texture.

CaMML: Context-Aware Multimodal Learner for Large Models

In this work, we introduce Context-Aware MultiModal Learner (CaMML), for tuning large multimodal models (LMMs). CaMML, a lightweight module, is crafted to seamlessly integrate multimodal contextual samples into large models, thereby empowering the model to derive knowledge from analogous, domain-specific, up-to-date information and make grounded inferences. Importantly, CaMML is highly scalable and can efficiently handle lengthy multimodal context examples owing to its hierarchical design. Based on CaMML, we have developed two multimodal models, CaMML-7B and CaMML-13B, that have shown exceptional performance across an array of benchmark datasets for multimodal tasks. Remarkably, CaMML-13B achieves the state-of-the-art performance on over ten widely recognized multimodal benchmark datasets, surpassing LLaVA-1.5 (13B) with a noticeable margin, without integration of any external resources. Moreover, we have conducted extensive ablative studies to inspect the inner workings of CaMML and performed qualitative analyses to showcase its effectiveness in handling real-world challenging cases.

USM-Lite: Quantization and Sparsity Aware Fine-tuning for Speech Recognition with Universal Speech Models

End-to-end automatic speech recognition (ASR) models have seen revolutionary quality gains with the recent development of large-scale universal speech models (USM). However, deploying these massive USMs is extremely expensive due to the enormous memory usage and computational cost. Therefore, model compression is an important research topic to fit USM-based ASR under budget in real-world scenarios. In this study, we propose a USM fine-tuning approach for ASR, with a low-bit quantization and N:M structured sparsity aware paradigm on the model weights, reducing the model complexity from parameter precision and matrix topology perspectives. We conducted extensive experiments with a 2-billion parameter USM on a large-scale voice search dataset to evaluate our proposed method. A series of ablation studies validate the effectiveness of up to int4 quantization and 2:4 sparsity. However, a single compression technique fails to recover the performance well under extreme setups including int2 quantization and 1:4 sparsity. By contrast, our proposed method can compress the model to have 9.4% of the size, at the cost of only 7.3% relative word error rate (WER) regressions. We also provided in-depth analyses on the results and discussions on the limitations and potential solutions, which would be valuable for future studies.

Labels Need Prompts Too: Mask Matching for Natural Language Understanding Tasks

Textual label names (descriptions) are typically semantically rich in many natural language understanding (NLU) tasks. In this paper, we incorporate the prompting methodology, which is widely used to enrich model input, into the label side for the first time. Specifically, we propose a Mask Matching method, which equips an input with a prompt and its label with another, and then makes predictions by matching their mask representations. We evaluate our method extensively on 8 NLU tasks with 14 datasets. The experimental results show that Mask Matching significantly outperforms its counterparts of fine-tuning and conventional prompt-tuning, setting up state-of-the-art performances in several datasets. Mask Matching is particularly good at handling NLU tasks with large label counts and informative label names. As pioneering efforts that investigate the label-side prompt, we also discuss open issues for future study.

WordScape: a Pipeline to extract multilingual, visually rich Documents with Layout Annotations from Web Crawl Data

We introduce WordScape, a novel pipeline for the creation of cross-disciplinary, multilingual corpora comprising millions of pages with annotations for document layout detection. Relating visual and textual items on document pages has gained further significance with the advent of multimodal models. Various approaches proved effective for visual question answering or layout segmentation. However, the interplay of text, tables, and visuals remains challenging for a variety of document understanding tasks. In particular, many models fail to generalize well to diverse domains and new languages due to insufficient availability of training data. WordScape addresses these limitations. Our automatic annotation pipeline parses the Open XML structure of Word documents obtained from the web, jointly providing layout-annotated document images and their textual representations. In turn, WordScape offers unique properties as it (1) leverages the ubiquity of the Word file format on the internet, (2) is readily accessible through the Common Crawl web corpus, (3) is adaptive to domain-specific documents, and (4) offers culturally and linguistically diverse document pages with natural semantic structure and high-quality text. Together with the pipeline, we will additionally release 9.5M urls to word documents which can be processed using WordScape to create a dataset of over 40M pages. Finally, we investigate the quality of text and layout annotations extracted by WordScape, assess the impact on document understanding benchmarks, and demonstrate that manual labeling costs can be substantially reduced.

CF-NeRF: Camera Parameter Free Neural Radiance Fields with Incremental Learning

Neural Radiance Fields (NeRF) have demonstrated impressive performance in novel view synthesis. However, NeRF and most of its variants still rely on traditional complex pipelines to provide extrinsic and intrinsic camera parameters, such as COLMAP. Recent works, like NeRFmm, BARF, and L2G-NeRF, directly treat camera parameters as learnable and estimate them through differential volume rendering. However, these methods work for forward-looking scenes with slight motions and fail to tackle the rotation scenario in practice. To overcome this limitation, we propose a novel \underline{c}amera parameter \underline{f}ree neural radiance field (CF-NeRF), which incrementally reconstructs 3D representations and recovers the camera parameters inspired by incremental structure from motion (SfM). Given a sequence of images, CF-NeRF estimates the camera parameters of images one by one and reconstructs the scene through initialization, implicit localization, and implicit optimization. To evaluate our method, we use a challenging real-world dataset NeRFBuster which provides 12 scenes under complex trajectories. Results demonstrate that CF-NeRF is robust to camera rotation and achieves state-of-the-art results without providing prior information and constraints.

Decoupling Degradation and Content Processing for Adverse Weather Image Restoration

Adverse weather image restoration strives to recover clear images from those affected by various weather types, such as rain, haze, and snow. Each weather type calls for a tailored degradation removal approach due to its unique impact on images. Conversely, content reconstruction can employ a uniform approach, as the underlying image content remains consistent. Although previous techniques can handle multiple weather types within a single network, they neglect the crucial distinction between these two processes, limiting the quality of restored images. This work introduces a novel adverse weather image restoration method, called DDCNet, which decouples the degradation removal and content reconstruction process at the feature level based on their channel statistics. Specifically, we exploit the unique advantages of the Fourier transform in both these two processes: (1) the degradation information is mainly located in the amplitude component of the Fourier domain, and (2) the Fourier domain contains global information. The former facilitates channel-dependent degradation removal operation, allowing the network to tailor responses to various adverse weather types; the latter, by integrating Fourier's global properties into channel-independent content features, enhances network capacity for consistent global content reconstruction. We further augment the degradation removal process with a degradation mapping loss function. Extensive experiments demonstrate our method achieves state-of-the-art performance in multiple adverse weather removal benchmarks.