Deep learning-based malicious traffic detection requires large-scale, accurately labeled datasets. However, realistic malicious traffic datasets contain noisy labels due to annotation errors, which can degrade the performance of detection models. Existing methods for handling noisy labels involve robust training or data cleaning. However, both approaches risk excluding hard samples from model update, leading to suboptimal decision boundaries and reduced classification performance. To this end, we introduce TRACE, a novel framework that leverages multi-loss optimization to induce robust traffic representations while utilizing all samples. TRACE utilizes a one-dimensional convolutional neural network (1D CNN) with multi-head attention to generate fine-grained traffic representations. Then, TRACE employs multi-loss optimization to enhance representation robustness, ensuring representation integrity and class separability, thereby laying the foundation for effective decision boundaries. The experimental results on two public datasets demonstrate that TRACE achieves an accuracy of 88.3% and an F1-score of 81% with 90% label noise, outperforming the state-of-the-art methods.

As many mobile apps generate substantial background traffic without active user interaction, network operators face increasing challenges in traffic management and analysis. However, existing approaches lack systematic methods for analyzing and identifying app background traffic. This paper presents BACKTRACKER, a novel background traffic identification system for mobile apps. To establish a reliable background traffic dataset, we design a semi-automated traffic collection framework integrating Android device with network traffic interception. We propose a URL similarity algorithm based on Levenshtein distance for accurate foreground-background traffic differentiation. Furthermore, we develop a hierarchical recognition model that combines statistical stability with deep learning expressiveness, integrating data augmentation, multi-head attention and BiLSTM networks for robust feature learning. Our extensive evaluation shows that BACKTRACKER significantly outperforms baseline methods, achieving 98.36% F1-score in background traffic identification. The results demonstrate BACKTRACKER’s effectiveness in background traffic analysis under encrypted network environments.

This paper presents a novel category agnostic model for visual rearrangement task, which can help an embodied agent to physically recover the shuffled scene configuration without any category concepts to the goal configuration. Previous methods usually follow a similar architecture, completing the rearrangement task by aligning the scene changes of the goal and shuffled configuration, according to the semantic scene graphs. However, constructing scene graphs requires the inference of category labels, which not only causes the accuracy drop of the entire task but also limits the application in real world scenario. In this paper, we delve deep into the essence of visual re-arrangement task and focus on the two most essential issues, scene change detection and scene change matching. We utilize the movement and the protrusion of point cloud to accurately identify the scene changes and match these changes depending on the similarity of category agnostic appearance feature. Moreover, to assist the agent to explore the environment more efficiently and comprehensively, we propose a closer-aligned-retrace exploration policy, aiming to observe more details of the scene at a closer distance. We conduct extensive experiments onAI2THOR Rearrangement Challenge based on RoomR dataset and a new multi-room multi-instance dataset MrMiR collected by us. The experimental results demonstrate the effectiveness of our proposed method.

Visual navigation is to let the agent reach the target according to the continuous visual input. In most previous works, visual navigation is usually assumed to be done in a static and ideal environment: the target is always reachable with no need to alter the environment. However, the “messy” environments are more general and practical in our daily lives, where the agent may get blocked by obstacles. Thus Interactive Navigation (InterNav) is introduced to navigate to the objects in more realistic “messy” environments according to the object interaction. Prior work on InterNav learns shortterm interaction through extensive trials with reinforcement learning. However, interaction does not guarantee efficient navigation, that is, plan-ning obstacle interactions that make shorter paths and con-sume less effort is also crucial. In this paper, we introduce an effect-oriented affordance map to enable longterm interactive navigation, extending the existing map-based nav-igation framework to the domain of dynamic environment. We train a set of affordance functions predicting available interactions and the time cost of removing obstacles, which informatively support an interactive modular system to ad-dress interaction and longterm planning. Experiments on the ProcTHOR simulator demonstrate the capability of our affordance-driven system in longterm navigation in complex dynamic environments.

We studied how robots could stop people from repeatedly obstructing them by using reactions that people commonly use. From 35 hours of observation of people in a shopping mall, we identified one commonly used reaction, which we named suggestive avoidance. It consists of making a quick movement to the side while rotating the body and gaze toward the obstructing person, in a way that seems to imply that they were bothered by the obstruction. We modeled the human suggestive avoidance behavior, implemented it on a robot, and tested it both in a lab experiment and a field study. The results from the lab study confirmed that people perceive a robot using suggestive avoidance as being more bothered, as well as more human-like. The field study showed that when a robot uses suggestive avoidance people are less likely to bother it again.

We developed an autonomous hatshop robot for encouraging customers to try on hats by providing comments that appropriately fit their actions, and in such a way also indirectly exerting social pressure. To enable it to offer such a service smoothly in a real shop, we developed a large system (around 150k lines of code with 23 ROS packages) integrated with various technologies, like people tracking, shopping activity recognition and navigation. The robot needed to move in narrow corridors, detect customers, and recognise their shopping activities. We employed an iterative development process, repeating trial-and-error integration with the robot in the actual shop, while also collecting real-world data during field-testing. This process enabled us to improve our shopping activity recognition system by collecting real-world data, and to adapt our software modules to the target shop environment. We report the lessons learnt during our system development process. The results of our 11-day field trial show that our robot was able to provide its services reasonably well. Many customers expressed a positive impression of the robot and its services.

We developed a motion control algorithm for a social mobile robot to intuitively convey its intent via social cues to pass through aisles and avoid misunderstanding in passing interactions with people, which frequently occur when a robot navigates in narrow shared environments. Inspired by observations of human behavior, the proposed algorithm estimates the extent to which a person understands the robot's intent on the basis of the person's reactions to the oncoming robot and provides the robot with corresponding motion strategies for effective passing interactions. We implemented the proposed algorithm onto an omni-directional humanoid robot and conducted a field study over six days in a store with 75 cm wide narrow aisles. The resulting behaviors of 50 customers demonstrated that our proposed method provided people with a clearer understanding of the robot's intent in passing interactions, and thus the robot had more opportunity (73.1%) to pass through aisles compared to 16.7% if the robot moved and then waited for people to make space.

Resistive random access memory (RRAM) is a promising technology for energy-efficient in-memory computing. However, due to technology limits, RRAM device faces a series of reliability issues. Deep neural network (DNN) computing based on RRAM suffers from accuracy degradation. On the one hand, offline DNN training solutions are difficult to fully consider and simulate all nonidealities. Worse still, new error or nonideality may come up with the usage of RRAM, which further deteriorates the effectiveness of offline training. On the other hand, online training poses great challenges on programming overhead and device lifetime. The iterative write-verify technique to program multi-bit RRAM cells prolongs write latency more than 10× longer than read latency. To overcome these issues, we propose a compensation architecture and a software and hardware co-training design to mitigate the realistic network accuracy loss in RRAM-based computing-in-memory chips. Firstly, we add trainable compensation channels in crossbars utilizing the residual resource after original weight mapping. Secondly, an offline training procedure with computing output from hardware is triggered to settle down appropriate weight value in compensation channels. Experimental results demonstrate that the proposed design can guarantee ≤ 0.8% loss of accuracy in DNN on MNIST and CIFAR10 dataset even when nonidealities reduce the original accuracy down to ≤73%.

Encrypted traffic classification using deep learning models based on packet length sequences has shown promising results. However, in real-world network conditions, network-induced phenomena such as packet loss, packet retransmission, and packet disorder are prevalent, leading to a decline in performance. To address this challenge, we propose Fractal, a novel approach designed to enhance existing deep learning models by integrating data augmentation and contrastive learning, thereby facilitating robust encrypted traffic classification under various network conditions. Specifically, Fractal employs three data augmentations to simulate different network conditions, generating diverse packet length sequences from the same flow. Contrastive learning is then leveraged to distill robust features from these augmented sequences. Fractal enables deep learning model to discern the intrinsic patterns of each flow, regardless of the variance in packet length sequences caused by network-induced phenomena. Our comprehensive evaluations demonstrate that Fractal enhances the classification performance of deep learning models under different network conditions, achieving 23% increase in accuracy and 15% improvement in F1-score.

We developed a security guard robot that is specifically designed to manage queues of people and conducted a field trial at an actual public event to assess its effectiveness. However, the acceptance of robot instructions or admonishments poses challenges in real-world applications. Our primary objective was to achieve an effective and socially acceptable queue-management solution. To accomplish this, we took inspiration from human security guards whose role has already been well received in society. Our robot, whose design embodied the image of a professional security guard, focused on three key aspects: duties, professional behavior, and appearance. To ensure its competence, we interviewed professional security guards to deepen our understanding of the responsibilities associated with queue management. Based on their insights, we incorporated features of ushering, admonishing, announcing, and question answering into the robot’s functionality. We also prioritized the modeling of professional ushering behavior. During a 10-day field trial at a children’s amusement event, we interviewed both the visitors who interacted with the robot and the event staff. The results revealed that visitors generally complied with its ushering and admonishments, indicating a positive reception. Both visitors and event staff expressed an overall favorable impression of the robot and its queue-management services. These findings suggest that our proposed security guard robot shows great promise as a solution for effective crowd handling in public spaces.

Service and social robot in public scenarios will face various tasks in future applications, such as guiding people or admonishing them to provide assistance or convey social norms. Robots in public spaces might also incorporate roles of authority figures who might admonish people (e.g., security or guard robots). However, recent investigations showed that people ignore the admonishment of robots. Thus, in this work, we are looking at the reasons why people might ignore robots based on the Cognitive Dissonance Theory (CDT). We present the results of two consecutive field observations where a robot admonishes participants (i.e., pedestrians in a shopping mall) and requests them to stop using a smartphone while walking, which is considered an unmoral behavior. In the first field observation, we approached 160 participants over four days and conducted semi-structured interviews with 19 of them. Approximately half of the people ignored the robot, and half of them followed the instructions. Our interview results show that people who ignore the robot indeed use trivialization as a cognitive dissonance reduction strategy to justify ignoring the robot. Based on our analysis of the results, we developed a counter-trivialization strategy that anticipates this dissonance reduction strategy. We admonished 167 participants in our second field observation over four days, and our results show that significantly fewer people ignore the instructions of the robot when the robot uses a counter-trivialization strategy.

Omni-directional robots have gradually been popular for social interactions with people in human environments. The characteristics of omni-directional bases allow the robots to change their body orientation freely while moving straight. However, human spectators show dislike when observing robots behave unnaturally. In this paper, we observed how humans naturally move to goals and then developed a motion planning algorithm for omni-directional robots to resemble human movements in a time-efficient manner. Instead of treating the translation and rotation of a robot separately, the proposed motion planner couples the two motions with constraints inspired from the observation of human behaviors. We implemented the proposed method onto an omni-directional robot and conducted navigation experiments in a shop with shelves and narrow corridors at width of 90cm. Results from a within-participants study of 300 human spectators validated that the proposed human-inspired motion planner provided people with more natural and predictable feelings compared to the common rotate-while-move or rotate-then-move strategies.

In this paper, we propose and experimentally verify a distributed formation control algorithm for a group of multirotor unmanned aerial vehicles (UAVs). The algorithm brings the whole group of UAVs simultaneously to a prescribed submanifold that determines the formation shape in an asymptotically stable fashion in two- and three-dimensional environments. The complete distributed control framework is implemented with the combination of a fast model predictive control method executed at 50 Hz on low-power computers onboard multirotor UAVs and validated via a series of hardware-in-the-loop simulations and real-robot experiments. The experiments are configured to study the control performance in various formation cases of arbitrary time-varying (e.g., expanding, shrinking, or moving) shapes. In the actual experiments, up to four multirotors have been implemented to form arbitrary triangular, rectangular, and circular shapes drawn by the operator via a human-robot interaction device. We also carry out hardware-in-the-loop simulations using up to six onboard computers to achieve spherical formations and a formation moving through obstacles.

TED Talks are short, powerful talks given by some of the world's brightest minds - scientists, philanthropists, businessmen, artists, and many others. Funded by members and advertising, these talks are free to access by the public on the TED website and TED YouTube channel, and many videos have become viral phenomena. In this research project, we perform a visual analysis of TED Talk videos and playlists to gain a good understanding of the trends and relationships between TED Talk topics.

This paper presents a nonlinear control approach for quadrotor Micro Aerial Vehicles (MAVs), which combines a backstepping-like regulator based on the solution of a certain class of global output regulation problems for the rigid body equations on SO(3), a robust controller for the system with bounded disturbances, as well as a trajectory generator using a model predictive control method. The proposed algorithm is endowed with strong convergence properties so that it allows the quadrotor MAVs to reach almost all the desired attitudes. The control approach is implemented on a high-payload-capable quadcopter with unstructured dynamics and unknown disturbances. The performance of our algorithm is demonstrated through a series of experimental evaluations and comparisons with another control method on normal and aggressive trajectory tracking tasks.