Source code is made available at https://github.com/ymLeiFDU/LICO.

Interpreting the decisions of deep learning models has been actively studied since the explosion of deep neural networks. One of the most convincing interpretation approaches is salience-based visual interpretation, such as Grad-CAM, where the generation of attention maps depends merely on categorical labels. Although existing interpretation methods can provide explainable decision clues, they often yield partial correspondence between image and saliency maps due to the limited discriminative information from one-hot labels. This paper develops a Language-Image COnsistency model for explainable image classification, termed LICO, by correlating learnable linguistic prompts with corresponding visual features in a coarse-to-fine manner. Specifically, we first establish a coarse global manifold structure alignment by minimizing the distance between the distributions of image and language features. We then achieve fine-grained saliency maps by applying optimal transport (OT) theory to assign local feature maps with class-specific prompts. Extensive experimental results on eight benchmark datasets demonstrate that the proposed LICO achieves a significant improvement in generating more explainable attention maps in conjunction with existing interpretation methods such as Grad-CAM. Remarkably, LICO improves the classification performance of existing models without introducing any computational overhead during inference.

Interpreting the decisions of deep learning models has been actively studied since the explosion of deep neural networks. One of the most convincing interpretation approaches is salience-based visual interpretation, such as Grad-CAM, where the generation of attention maps depends merely on categorical labels. Although existing interpretation methods can provide explainable decision clues, they often yield partial correspondence between image and saliency maps due to the limited discriminative information from one-hot labels. This paper develops a Language-Image COnsistency model for explainable image classification, termed LICO, by correlating learnable linguistic prompts with corresponding visual features in a coarse-to-fine manner. Specifically, we first establish a coarse global manifold structure alignment by minimizing the distance between the distributions of image and language features.

Optimizing black-box functions is a fundamental problem in science and engineering. To solve this problem, many approaches learn a surrogate function that estimates the underlying objective from limited historical evaluations. Large Language Models (LLMs), with their strong pattern-matching capabilities via pretraining on vast amounts of data, stand out as a potential candidate for surrogate modeling. However, directly prompting a pretrained language model to produce predictions is not feasible in many scientific domains due to the scarcity of domain-specific data in the pretraining corpora and the challenges of articulating complex problems in natural language. In this work, we introduce LICO, a general-purpose model that extends arbitrary base LLMs for black-box optimization, with a particular application to the molecular domain. To achieve this, we equip the language model with a separate embedding layer and prediction layer, and train the model to perform in-context predictions on a diverse set of functions defined over the domain. Once trained, LICO can generalize to unseen molecule properties simply via in-context prompting. LICO achieves state-of-the-art performance on PMO, a challenging molecular optimization benchmark comprising over 20 objective functions.

The democratization of any effective technology happens automatically by virtue of its success, even if the complexity it presents initially overwhelms some of the smartest people who wield it. But after six decades of commercial computing in the datacenter, we have certainly learned a thing or two about helping this process of adoption and integration along. There's nothing intrinsically special about artificial intelligence (AI) in this regard, which is arguably just the latest evolution in a long line of sophisticated data processing tools. Mainframes were kept in glasshouses as a kind of temple of computing during the 1960s and 1970s before being mimicked and copied into minicomputers. Eventually, PCs spawned two seismic shifts: becoming powerful enough to be servers and run mission-critical workloads in a client-server environment.

We hear about new smart technology all the time, from smart speakers to smart watches and even smart refrigerators. While smart products promise to make our lives better, how do we make them truly complement our lives? Beyond just personal and consumer devices, we now have more connected devices in our factories and businesses producing vast amounts of data from a burgeoning number of IoT devices. Such devices, including sensors, cameras, and wearables could be the key to making our workplaces "smarter." However, we must carefully leverage and interpret the vast amounts of data--if not, we risk data becoming a cost burden, adding more noise instead of providing insight and order.

This sponsored post from Lenovo's Bhushan Desam covers how new HPC tools like Lenovo's LiCO (Lenovo Intelligent Computing Orchestration) are working to address the growing popularity of AI and to simplify the convergence of HPC and AI. Artificial Intelligence (AI) is coming for your HPC cluster – and while there are no autonomous robots taking over the data center, some days it might feel that way to cluster administrators. The HPC cluster looks very attractive to the "outside world", particularly to those who will need performance beyond a single system or workstation. That is, until they try to use it and realize there is a learning curve they have to overcome. AI workloads are well suited for running on a cluster – but is your cluster management ready for AI users?