Semi-supervised medical image segmentation has advanced significantly by utilizing pseudo-labeled annotations. However, ensuring pseudo-label accuracy remains challenging, often causing misclassification and confirmation bias. Existing methods mainly use prediction uncertainty to exclude or downweight uncertain regions, but these areas frequently coincide with diagnostically important zones, such as lesion cores or tissue boundaries. Neglecting them can thus degrade segmentation performance. To address this, we propose a Dynamic Contrastive Correction Network (DCCN) that corrects uncertain regions instead of ignoring them. DCCN aligns features from high-uncertainty areas with dynamically assigned classes via contrastive learning, reconstructing the uncertain feature space. Additionally, a Multi-layer Sampling (MLS) module leverages boundary-aware sampling to focus contrastive learning on uncertain tissue boundaries. Experiments on two public datasets show that DCCN surpasses previous SOTA methods and effectively mitigates the challenges of high-uncertainty regions.

Photoplethysmography (PPG) has been widely used in consumer and medical devices for assessing heart rate blood oxygen and blood pressure levels, together with electrocardiography (ECG). One of the major challenges is recording PPG with fast changing ambient light and motion artefacts that generating a changing baseline of the signal. Light to digital converters (LDCs) proposed in recent years show excellent resolution and power efficiency, however it may get saturated during fast ambient and motion events. Existing chopping ambient light removal and hysteresis DC removal techniques cannot track fast-changing components. Therefore, this paper proposes a current domain level crossing-based technique to assist a dual slope LDC, including a dynamic biased low-power continuous time comparator. The PPG readout frontend is implemented in a 55nm standard CMOS process and consumes 12.4μW-56.6μW depending on the ambient light intensity. It achieves a dynamic range of 102dB and an SNDR of 75dB at AC signal path.

Deep learning techniques have advanced rapidly in recent years, leading to significant progress in pre-trained and fine-tuned large-scale AI models. For example, in the natural language processing domain, the traditional "pre-train, fine-tune" paradigm is shifting towards the "pre-train, prompt, and predict" paradigm, which has achieved great success on many tasks across different application domains such as ChatGPT/BARD for Conversational AI and P5 for a unified recommendation system. Moreover, there has been a growing interest in models that combine vision and language modalities (vision-language models) which are applied to tasks like Visual Captioning/Generation. Considering the recent technological revolution, it is essential to emphasize these paradigm shifts and highlight the paradigms with the potential to solve different tasks. We thus provide a platform for academic and industrial researchers to showcase their latest work, share research ideas, discuss various challenges, and identify areas where further research is needed in pre-training, fine-tuning, and prompt-learning methods for large-scale AI models. We foster the development of a strong research community focused on solving challenges related to large-scale AI models, providing superior and impactful strategies that can change people's lives in the future.

In the current digital world, web search engines and recommendation systems are continuously evolving, opening up new potential challenges every day which require more sophisticated and efficient data mining and machine learning solutions to satisfy the needs of sellers and consumers as well as marketers. The quality of search and recommendation systems impacts customer retention, time on site, and sales volume. For instance, with often sparse conversion rates, highly personalized contents, heterogeneous digital sources, more rigorous and effective models are required to be developed by research engineers and data scientists. At the same time, deep learning has started to show great impact in many industrial applications which are capable of processing complicated, large-scale and real-time data. Deep learning not only provides more opportunities to increase conversion rates and improve revenue through a positive customer experience, but also provides customers with personalized contents along with their personal shopping journey. Due to this rapid growth of the digital world, there is a need to bring professionals together from both academic research and the industry to solve real-world problems. This workshop fosters the development of a strong research community focused on solving deep learning based large-scale web search, personalized search, recommendation and ranking relevance problems that provide superior digital experience to all users.

Time series data is ubiquitous, and accurate time series forecasting is vital for many real-world application domains, including retail, healthcare, supply chain, climate science, e-commerce and economics. Forecasting, in general, has led to broad impact and a diverse range of applications. However, with large-scale, high-dimensional time-series data available, more advanced techniques must be invented or improved for highly accurate predictions. Latest data mining and machine learning techniques play a crucial role in the next generation of forecasting models. In this Applied Machine Learning Methods for Time Series Forecasting (AMLTS) workshop, we focus on effective and accurate latest machine learning approaches to solve various real-world problems. With this workshop's ability to attract audiences across various domains, we invite experienced industrial practitioners and researchers to help uncover new approaches and break new ground in time-series modelings' challenging and vital settings.

An average consumer spends 8+ hours a day across all devices interacting with online content almost entirely sponsored by advertisements. At over $450B global market size in 2022 and expected to pass $1T by 2027, online advertising has already surpassed traditional ads in global spend. Moreover, computational advertising in particular is perhaps the most visible and ubiquitous application of machine learning and one that interacts directly with consumers. When done right, ads help us enrich our lives and creep us out when done badly. Looking at the published literature over the last few years, many researchers might consider computational advertising as a mature field. Yet, the opposite is true. The field is evolving, however, from ads controlled by monolithic publishers and randomly rotating banner ads to highly personalized content experiences in news feeds on mobile devices and even on TV-all utilizing data amassed from petabytes of stored user data. Ads are far from done.

Building recommendation systems for enterprise software has many unique challenges that are different from consumer-facing systems. When applied to different organizations, the data used to power those recommendation systems vary substantially in both quality and quantity due to differences in their operational practices, marketing strategies, and targeted audiences. At Salesforce, as a cloud provider of such a system with data across many different organizations, naturally, it makes sense to pool data from different organizations to build a model that combines all values from different brands. However, multiple issues like how do we make sure a model trained with pooled data can still capture customer specific characteristics, how do we design the system to handle those data responsibly and ethically, i.e., respecting contractual agreements with our clients, legal and compliance requirements, and the privacy of all the consumers. In this proposal, We present a framework that not only utilizes enriched user-level data across organizations, but also boosts business-specific characteristics in generating personal recommendations. We will also walk through key privacy considerations when designing such a system.

Next best action (NBA) is a technique that is widely considered as the best practice in modern personalized marketing. It takes users' unique characteristics into consideration and recommends next actions that help users progress towards business goals as quickly and smoothly as possible. Many NBA engines are built with rules handcrafted by marketers based on experience or gut feelings. It is not effective. In this proposal, we show our machine learning based approach for such a real-time recommendation engine, detail our design choices, and discuss evaluation techniques.

Trajectory sharing and searching have received significant attentions in recent years. In this paper, we propose and investigate a novel problem called User Oriented Trajectory Search (UOTS) for trip recommendation. In contrast to conventional trajectory search by locations (spatial domain only), we consider both spatial and textual domains in the new UOTS query. Given a trajectory data set, the query input contains a set of intended places given by the traveler and a set of textual attributes describing the traveler's preference. If a trajectory is connecting/close to the specified query locations, and the textual attributes of the trajectory are similar to the traveler'e preference, it will be recommended to the traveler for reference. This type of queries can bring significant benefits to travelers in many popular applications such as trip planning and recommendation.

Shortest path query is one of the most fundamental queries in spatial network databases. There exist algorithms that can process shortest path queries in real time. However, many complex applications require more than just the calculation of a single shortest path. For example, one of the common ways to determine the importance (or price) of a vertex or an edge in spatial network is to use Vickrey pricing, which intuitively values the vertex v (or edge e ) based on how much harder for travelling from the sources to the destinations without using v (or e ). In such cases, the alternative shortest paths without using v (or e ) are required. In this article, we propose using a precomputation based approach for both single pair alternative shortest path and all pairs shortest paths processing. To compute the alternative shortest path between a source and a destination efficiently, a naïive way is to precompute and store all alternative shortest paths between every pair of vertices avoiding every possible vertex (or edge), which requires O ( n 4 ) space. Currently, the state of the art approach for reducing the storage cost is to choose a subset of the vertices as center points, and only store the single-source alternative shortest paths from those center points. Such approach has the space complexity of O ( n 2 log n ). We propose a storage scheme termed iSPQF , which utilizes shortest path quadtrees by observing the relationships between each avoiding vertex and its corresponding alternative shortest paths. We have reduced the space complexity from the naïive O ( n 4 ) (or the state of the art O ( n 4 log n )) to O (min( γ, L ) n 1.5 ) with comparable query performance of O ( K ), where K is the number of vertices in the returned paths, L is the diameter of the spatial network, and γ is a value that depends on the structure of the spatial network, which is empirically estimated to be 40 for real road networks. Experiments on real road networks have shown that the space cost of the proposed iSPQF is scalable, and both the algorithms based on iSPQF are efficient.

In this paper, we propose and investigate a novel spatial query called Reverse Path Nearest Neighbor (R-PNN) search to find the most accessible locations in road networks. Given a trajectory data-set and a list of location candidates specified by users, if a location o is the Path Nearest Neighbor (PNN) of k trajectories, the influence-factor of o is defined as k and the R-PNN query returns the location with the highest influence-factor. The R-PNN query is an extension of the conventional Reverse Nearest Neighbor (RNN) search. It can be found in many important applications such as urban planning, facility allocation, traffic monitoring, etc. To answer the R-PNN query efficiently, an effective trajectory data pre-processing technique is conducted in the first place. We cluster the trajectories into several groups according to their distribution. Based on the grouped trajectory data, a two-phase solution is applied. First, we specify a tight search range over the trajectory and location data-sets. The efficiency study reveals that our approach defines the minimum search area. Second, a series of optimization techniques are adopted to search the exact PNN for trajectories in the candidate set. By combining the PNN query results, we can retrieve the most accessible locations. The complexity analysis shows that our solution is optimal in terms of time cost. The performance of the proposed R-PNN query processing is verified by extensive experiments based on real and synthetic trajectory data in road networks.

A point detour is a temporary deviation from a user preferred path P (not necessarily a shortest network path) for visiting a data point such as a supermarket or McDonald's. The goodness of a point detour can be measured by the additional traveling introduced, called point detour cost or simply detour cost. Given a preferred path to be traveling on, Best Point Detour (BPD) query aims to identify the point detour with the minimum detour cost. This problem can be frequently found in our daily life but is less studied. In this work, the efficient processing of BPD query is investigated with support of devised optimization techniques. Furthermore, we investigate continuous-BPD query with target at the scenario where the path to be traveling on continuously changes when a user is moving to the destination along the preferred path. The challenge of continuous-BPD query lies in finding a set of update locations which split P into partitions. In the same partition, the user has the same BPD. We process continuous-BPD query by running BPD queries in a deliberately planned strategy. The efficiency study reveals that the number of BPD queries executed is optimal. The efficiency of BPD query and continuous-BPD query processing has been verified by extensive experiments.