Large Multimodal Models (LMMs) have emerged as powerful models capable of understanding various data modalities, including text, images, and videos. LMMs encode both text and visual data into tokens that are then combined and processed by an integrated Large Language Model (LLM). Including visual tokens substantially increases the total token count, often by thousands. The increased input length for LLM significantly raises the complexity of inference, resulting in high latency in LMMs. To address this issue, token pruning methods, which remove part of the visual tokens, are proposed. The existing token pruning methods either require extensive calibration and fine-tuning or rely on suboptimal importance metrics which results in increased redundancy among the retained tokens. In this paper, we first formulate token pruning as Max-Min Diversity Problem (MMDP) where the goal is to select a subset such that the diversity among the selected tokens is maximized. Then, we solve the MMDP to obtain the selected subset and prune the rest. The proposed method, DivPrune, reduces redundancy and achieves the highest diversity of the selected tokens. By ensuring high diversity, the selected tokens better represent the original tokens, enabling effective performance even at high pruning ratios without requiring fine-tuning. Extensive experiments with various LMMs show that DivPrune achieves state-of-the-art accuracy over 16 image- and video-language datasets. Additionally, DivPrune reduces both the end-to-end latency and GPU memory usage for the tested models. The code is available here⋄.

Curating high-quality datasets requires significant resources, making ownership protection crucial. Recently, to protect the ownership of an image dataset, imperceptible watermarking is used to store ownership information (i.e., watermark) into the individual samples. However, embedding the entire watermark into all samples causes redundancy, damaging dataset quality and extraction accuracy. In this paper, a multi-segment encoding-decoding method for dataset watermarking (called AMUSE) is proposed to adaptively map the original watermark into a set of shorter sub-messages and vice versa. The message encoder adjusts sub-message lengths based on protection requirements, and the decoder reconstructs the original message from the extracted sub-messages. AMUSE is plug-and-play and orthogonal to existing image watermarking methods. Extensive experiments with multiple watermarking solutions demonstrate that AMUSE improves the dataset quality and message extraction accuracy. The code is available in https://github.com/vbdi/amuse.

Building multi-modal language models has been a trend in the recent years, where additional modalities such as image, video, speech, etc. are jointly learned along with natural languages (i.e., textual information).Despite the success of these multi-modal language models with different modalities, there is no existing solution for neural network architectures and natural languages.Providing neural architectural information as a new modality allows us to provide fast architecture-2-text and text-2-architecture retrieval/generation services on the cloud with a single inference.Such solution is valuable in terms of helping beginner and intermediate ML users to come up with better neural architectures or AutoML approaches with a simple text query.In this paper, we propose ArchBERT, a bi-modal model for joint learning and understanding of neural architectures and natural languages, which opens up new avenues for research in this area.We also introduce a pre-training strategy named Masked Architecture Modeling (MAM) for a more generalized joint learning.Moreover, we introduce and publicly release two new bi-modal datasets for training and validating our methods.The ArchBERT's performance is verified through a set of numerical experiments on different downstream tasks such as architecture-oriented reasoning, question answering, and captioning (summarization).Datasets, codes, and demos are available as supplementary materials 1 .

In today's digital world, enterprises and individuals are generating massive data that is potentially useful for many data consumers with data driven applications. The emergence of data marketplaces is a step toward helping the data owners to monetize their digital assets and get connected to the potential buyers. The current data marketplaces cannot handle the challenges related to data ownership claims, illegal redistribution, and data ownership traceability. To overcome these problems in a general-purpose market, we propose a marketplace based on watermarking and Non-Fungible Token (NFT) technologies. In the proposed NFT-based marketplace, the owner's data is stored as an NFT where the underlying content of the NFT holds the watermarked data. The watermarked data is obtained by embedding some information about the owners and the buyers into the original data. The embedded information can later be extracted to identify the owner and the buyer of the traded data. Furthermore, the transactions corresponding to the NFT provide verifiable ownership proof and traceable ownership history. A Proof-Of-Concept (POC) implementation of the proposed marketplace that will be integrated within AI-Gallery Data Marketplace service in Huawei Cloud is presented for trading image data. An extensive set of experiments to measure the gas consumption on the blockchain and evaluate the robustness of the watermarked assets against 51 attacks are performed. Finally, a method based on error correction codes is proposed for improving the watermarking robustness in the implemented marketplace. The link for the codes and the POC demo is provided in the appendix.

Learning-based compression systems have shown great potential for multi-task inference from their latent-space representation of the input image. In such systems, the decoder is supposed to be able to perform various analyses of the input image, such as object detection or segmentation, besides decoding the image. At the same time, privacy concerns around visual ana-lytics have grown in response to the increasing capabilities of such systems to reveal private information. In this paper, we propose a method to make latent-space inference more privacy-friendly using mutual information-based criteria. In particular, we show how organizing and compressing the latent representation of the image according to task-specific mutual information can make the model maintain high analytics accuracy while becoming less able to reconstruct the input image and thereby reveal private information.

In recent studies, collaborative intelligence (CI) has emerged as a promising framework for deployment of Artificial Intelligence (AI)-based services on mobile/edge devices. In CI, the AI model (a deep neural network) is split between the edge and the cloud, and intermediate features are sent from the edge sub-model to the cloud sub-model. In this article, we study bit allocation for feature coding in multi-stream CI systems. We model task distortion as a function of rate using convex surfaces similar to those found in distortion-rate theory. Using such models, we are able to provide closed-form bit allocation solutions for single-task systems and scalarized multi-task systems. Moreover, we provide analytical characterization of the full Pareto set for 2-stream k -task systems, and bounds on the Pareto set for 3-stream 2-task systems. Analytical results are examined on a variety of DNN models from the literature to demonstrate wide applicability of the results.

Recent studies have shown that collaborative intelligence (CI) is a promising framework for deployment of Artificial Intelligence (AI)-based services on mobile devices. In CI, a deep neural network is split between the mobile device and the cloud. Deep features obtained at the mobile are compressed and transferred to the cloud to complete the inference. So far, the methods in the literature focused on transferring a single deep feature tensor from the mobile to the cloud. Such methods are not applicable to some recent, high-performance networks with multiple branches and skip connections. In this paper, we propose the first bit allocation method for multi-stream, multi-task CI. We first establish a model for the joint distortion of the multiple tasks as a function of the bit rates assigned to different deep feature tensors. Then, using the proposed model, we solve the rate-distortion optimization problem under a total rate constraint to obtain the best rate allocation among the tensors to be transferred. Experimental results illustrate the efficacy of the proposed scheme compared to several alternative bit allocation methods.

A promising way to deploy Artificial Intelligence (AI)-based services on mobile devices is to run a part of the AI model (a deep neural network) on the mobile itself, and the rest in the cloud. This is sometimes referred to as collaborative intelligence. In this framework, intermediate features from the deep network need to be transmitted to the cloud for further processing. We study the case where such features are used for multiple purposes in the cloud (multi-tasking) and where they need to be compressible in order to allow efficient transmission to the cloud. To this end, we introduce a new loss function that encourages feature compressibility while improving system performance on multiple tasks. Experimental results show that with the compression-friendly loss, one can achieve around 20% bitrate reduction without sacrificing the performance on several vision-related tasks.

Finding faces in images is one of the most important tasks in computer vision, with applications in biometrics, surveillance, human-computer interaction, and other areas. In our earlier work, we demonstrated that it is possible to tell whether or not an image contains a face by only examining the HEVC syntax, without fully reconstructing the image. In the present work we move further in this direction by showing how to localize faces in HEVC-coded images, without full reconstruction. We also demonstrate the benefits that such approach can have in privacy-friendly face localization.

Object tracking is the cornerstone of many visual analytics systems. While considerable progress has been made in this area in recent years, robust, efficient, and accurate tracking in real-world video remains a challenge. In this paper, we present a hybrid tracker that leverages motion information from the compressed video stream and a general-purpose semantic object detector acting on decoded frames to construct a fast and efficient tracking engine. The proposed approach is compared with several well-known recent trackers on the OTB tracking dataset. The results indicate advantages of the proposed method in terms of speed and/or accuracy. Other desirable features of the proposed method are its simplicity and deployment efficiency, which stems from the fact that it reuses the resources and information that may already exist in the system for other reasons.