Ladislau Bölöni

Jiaqi Xue, Mengxin Zheng, Ting Hua et al.

Large Language Models (LLMs) are progressively being utilized as machine learning services and interface tools for various applications. However, the security implications of LLMs, particularly in relation to adversarial and Trojan attacks, remain insufficiently examined. In this paper, we propose TrojLLM, an automatic and black-box framework to effectively generate universal and stealthy triggers. When these triggers are incorporated into the input data, the LLMs' outputs can be maliciously manipulated. Moreover, the framework also supports embedding Trojans within discrete prompts, enhancing the overall effectiveness and precision of the triggers' attacks. Specifically, we propose a trigger discovery algorithm for generating universal triggers for various inputs by querying victim LLM-based APIs using few-shot data samples. Furthermore, we introduce a novel progressive Trojan poisoning algorithm designed to generate poisoned prompts that retain efficacy and transferability across a diverse range of models. Our experiments and results demonstrate TrojLLM's capacity to effectively insert Trojans into text prompts in real-world black-box LLM APIs including GPT-3.5 and GPT-4, while maintaining exceptional performance on clean test sets. Our work sheds light on the potential security risks in current models and offers a potential defensive approach. The source code of TrojLLM is available at https://github.com/UCF-ML-Research/TrojLLM.

Saad Almohaimeed, Saleh Almohaimeed, Ashfaq Ali Shafin et al.

Detecting harmful content on social media, such as Twitter, is made difficult by the fact that the seemingly simple yes/no classification conceals a significant amount of complexity. Unfortunately, while several datasets have been collected for training classifiers in hate and offensive speech, there is a scarcity of datasets labeled with a finer granularity of target classes and specific targets. In this paper, we introduce THOS, a dataset of 8.3k tweets manually labeled with fine-grained annotations about the target of the message. We demonstrate that this dataset makes it feasible to train classifiers, based on Large Language Models, to perform classification at this level of granularity.

Jason Gardner, Ayan Dutta, Swapnoneel Roy et al.

The growing computational demands of deep reinforcement learning (DRL) have raised concerns about the environmental and economic costs of training large-scale models. While algorithmic efficiency in terms of learning performance has been extensively studied, the energy requirements, greenhouse gas emissions, and monetary costs of DRL algorithms remain largely unexplored. In this work, we present a systematic benchmarking study of the energy consumption of seven state-of-the-art DRL algorithms, namely DQN, TRPO, A2C, ARS, PPO, RecurrentPPO, and QR-DQN, implemented using Stable Baselines. Each algorithm was trained for one million steps each on ten Atari 2600 games, and power consumption was measured in real-time to estimate total energy usage, CO2-Equivalent emissions, and electricity cost based on the U.S. national average electricity price. Our results reveal substantial variation in energy efficiency and training cost across algorithms, with some achieving comparable performance while consuming up to 24% less energy (ARS vs. DQN), emitting nearly 68% less CO2, and incurring almost 68% lower monetary cost (QR-DQN vs. RecurrentPPO) than less efficient counterparts. We further analyze the trade-offs between learning performance, training time, energy use, and financial cost, highlighting cases where algorithmic choices can mitigate environmental and economic impact without sacrificing learning performance. This study provides actionable insights for developing energy-aware and cost-efficient DRL practices and establishes a foundation for incorporating sustainability considerations into future algorithmic design and evaluation.

Sharare Zehtabian, Siavash Khodadadeh, Damla Turgut et al.

Throughout the Covid-19 pandemic, a significant amount of effort had been put into developing techniques that predict the number of infections under various assumptions about the public policy and non-pharmaceutical interventions. While both the available data and the sophistication of the AI models and available computing power exceed what was available in previous years, the overall success of prediction approaches was very limited. In this paper, we start from prediction algorithms proposed for XPrize Pandemic Response Challenge and consider several directions that might allow their improvement. Then, we investigate their performance over medium-term predictions extending over several months. We find that augmenting the algorithms with additional information about the culture of the modeled region, incorporating traditional compartmental models and up-to-date deep learning architectures can improve the performance for short term predictions, the accuracy of medium-term predictions is still very low and a significant amount of future research is needed to make such models a reliable component of a public policy toolbox.

Sharare Zehtabian, Siavash Khodadadeh, Ladislau Bölöni et al.

The daily activities performed by a disabled or elderly person can be monitored by a smart environment, and the acquired data can be used to learn a predictive model of user behavior. To speed up the learning, several researchers designed collaborative learning systems that use data from multiple users. However, disclosing the daily activities of an elderly or disabled user raises privacy concerns. In this paper, we use state-of-the-art deep neural network-based techniques to learn predictive human activity models in the local, centralized, and federated learning settings. A novel aspect of our work is that we carefully track the temporal evolution of the data available to the learner and the data shared by the user. In contrast to previous work where users shared all their data with the centralized learner, we consider users that aim to preserve their privacy. Thus, they choose between approaches in order to achieve their goals of predictive accuracy while minimizing the shared data. To help users make decisions before disclosing any data, we use machine learning to predict the degree to which a user would benefit from collaborative learning. We validate our approaches on real-world data.

Hassam Ullah Sheikh, Ladislau Bölöni

The classic DQN algorithm is limited by the overestimation bias of the learned Q-function. Subsequent algorithms have proposed techniques to reduce this problem, without fully eliminating it. Recently, the Maxmin and Ensemble Q-learning algorithms have used different estimates provided by the ensembles of learners to reduce the overestimation bias. Unfortunately, these learners can converge to the same point in the parametric or representation space, falling back to the classic single neural network DQN. In this paper, we describe a regularization technique to maximize ensemble diversity in these algorithms. We propose and compare five regularization functions inspired from economics theory and consensus optimization. We show that the regularized approach significantly outperforms the Maxmin and Ensemble Q-learning algorithms as well as non-ensemble baselines.

Siavash Khodadadeh, Sharare Zehtabian, Saeed Vahidian et al.

Unsupervised meta-learning approaches rely on synthetic meta-tasks that are created using techniques such as random selection, clustering and/or augmentation. Unfortunately, clustering and augmentation are domain-dependent, and thus they require either manual tweaking or expensive learning. In this work, we describe an approach that generates meta-tasks using generative models. A critical component is a novel approach of sampling from the latent space that generates objects grouped into synthetic classes forming the training and validation data of a meta-task. We find that the proposed approach, LAtent Space Interpolation Unsupervised Meta-learning (LASIUM), outperforms or is competitive with current unsupervised learning baselines on few-shot classification tasks on the most widely used benchmark datasets. In addition, the approach promises to be applicable without manual tweaking over a wider range of domains than previous approaches.

Hassam Ullah Sheikh, Ladislau Bölöni

Many cooperative multi-agent problems require agents to learn individual tasks while contributing to the collective success of the group. This is a challenging task for current state-of-the-art multi-agent reinforcement algorithms that are designed to either maximize the global reward of the team or the individual local rewards. The problem is exacerbated when either of the rewards is sparse leading to unstable learning. To address this problem, we present Decomposed Multi-Agent Deep Deterministic Policy Gradient (DE-MADDPG): a novel cooperative multi-agent reinforcement learning framework that simultaneously learns to maximize the global and local rewards. We evaluate our solution on the challenging defensive escort team problem and show that our solution achieves a significantly better and more stable performance than the direct adaptation of the MADDPG algorithm.

Pooya Abolghasemi, Ladislau Bölöni

Recent research demonstrated that it is feasible to end-to-end train multi-task deep visuomotor policies for robotic manipulation using variations of learning from demonstration (LfD) and reinforcement learning (RL). In this paper, we extend the capabilities of end-to-end LfD architectures to object manipulation in clutter. We start by introducing a data augmentation procedure called Accept Synthetic Objects as Real (ASOR). Using ASOR we develop two network architectures: implicit attention ASOR-IA and explicit attention ASOR-EA. Both architectures use the same training data (demonstrations in uncluttered environments) as previous approaches. Experimental results show that ASOR-IA and ASOR-EA succeed ina significant fraction of trials in cluttered environments where previous approaches never succeed. In addition, we find that both ASOR-IA and ASOR-EA outperform previous approaches even in uncluttered environments, with ASOR-EA performing better even in clutter compared to the previous best baseline in an uncluttered environment.

Hassam Ullah Sheikh, Ladislau Bölöni

We explore a collaborative and cooperative multi-agent reinforcement learning setting where a team of reinforcement learning agents attempt to solve a single cooperative task in a multi-scenario setting. We propose a novel multi-agent reinforcement learning algorithm inspired by universal value function approximators that not only generalizes over state space but also over a set of different scenarios. Additionally, to prove our claim, we are introducing a challenging 2D multi-agent urban security environment where the learning agents are trying to protect a person from nearby bystanders in a variety of scenarios. Our study shows that state-of-the-art multi-agent reinforcement learning algorithms fail to generalize a single task over multiple scenarios while our proposed solution works equally well as scenario-dependent policies.

Hassam Ullah Sheikh, Ladislau Bölöni

We are considering a scenario where a team of bodyguard robots provides physical protection to a VIP in a crowded public space. We use deep reinforcement learning to learn the policy to be followed by the robots. As the robot bodyguards need to follow several difficult-to-reconcile goals, we study several primitive and composite reward functions and their impact on the overall behavior of the robotic bodyguards.

Siavash Khodadadeh, Ladislau Bölöni, Mubarak Shah

Few-shot or one-shot learning of classifiers requires a significant inductive bias towards the type of task to be learned. One way to acquire this is by meta-learning on tasks similar to the target task. In this paper, we propose UMTRA, an algorithm that performs unsupervised, model-agnostic meta-learning for classification tasks. The meta-learning step of UMTRA is performed on a flat collection of unlabeled images. While we assume that these images can be grouped into a diverse set of classes and are relevant to the target task, no explicit information about the classes or any labels are needed. UMTRA uses random sampling and augmentation to create synthetic training tasks for meta-learning phase. Labels are only needed at the final target task learning step, and they can be as little as one sample per class. On the Omniglot and Mini-Imagenet few-shot learning benchmarks, UMTRA outperforms every tested approach based on unsupervised learning of representations, while alternating for the best performance with the recent CACTUs algorithm. Compared to supervised model-agnostic meta-learning approaches, UMTRA trades off some classification accuracy for a reduction in the required labels of several orders of magnitude.

Pooya Abolghasemi, Amir Mazaheri, Mubarak Shah et al.

Several recent studies have demonstrated the promise of deep visuomotor policies for robot manipulator control. Despite impressive progress, these systems are known to be vulnerable to physical disturbances, such as accidental or adversarial bumps that make them drop the manipulated object. They also tend to be distracted by visual disturbances such as objects moving in the robot's field of view, even if the disturbance does not physically prevent the execution of the task. In this paper, we propose an approach for augmenting a deep visuomotor policy trained through demonstrations with Task Focused visual Attention (TFA). The manipulation task is specified with a natural language text such as `move the red bowl to the left'. This allows the visual attention component to concentrate on the current object that the robot needs to manipulate. We show that even in benign environments, the TFA allows the policy to consistently outperform a variant with no attention mechanism. More importantly, the new policy is significantly more robust: it regularly recovers from severe physical disturbances (such as bumps causing it to drop the object) from which the baseline policy, i.e. with no visual attention, almost never recovers. In addition, we show that the proposed policy performs correctly in the presence of a wide class of visual disturbances, exhibiting a behavior reminiscent of human selective visual attention experiments. Our proposed approach consists of a VAE-GAN network which encodes the visual input and feeds it to a Motor network that moves the robot joints. Also, our approach benefits from a teacher network for the TFA that leverages textual input command to robustify the visual encoder against various types of disturbances.

Rouhollah Rahmatizadeh, Pooya Abolghasemi, Ladislau Bölöni et al.

We propose a technique for multi-task learning from demonstration that trains the controller of a low-cost robotic arm to accomplish several complex picking and placing tasks, as well as non-prehensile manipulation. The controller is a recurrent neural network using raw images as input and generating robot arm trajectories, with the parameters shared across the tasks. The controller also combines VAE-GAN-based reconstruction with autoregressive multimodal action prediction. Our results demonstrate that it is possible to learn complex manipulation tasks, such as picking up a towel, wiping an object, and depositing the towel to its previous position, entirely from raw images with direct behavior cloning. We show that weight sharing and reconstruction-based regularization substantially improve generalization and robustness, and training on multiple tasks simultaneously increases the success rate on all tasks.

Jun Xu, Gurkan Solmaz, Rouhollah Rahmatizadeh et al.

In animal monitoring applications, both animal detection and their movement prediction are major tasks. While a variety of animal monitoring strategies exist, most of them rely on mounting devices. However, in real world, it is difficult to find these animals and install mounting devices. In this paper, we propose an animal monitoring application by utilizing wireless sensor networks (WSNs) and unmanned aerial vehicle (UAV). The objective of the application is to detect locations of endangered species in large-scale wildlife areas and monitor movement of animals without any attached devices. In this application, sensors deployed throughout the observation area are responsible for gathering animal information. The UAV flies above the observation area and collects the information from sensors. To achieve the information efficiently, we propose a path planning approach for the UAV based on a Markov decision process (MDP) model. The UAV receives a certain amount of reward from an area if some animals are detected at that location. We solve the MDP using Q-learning such that the UAV prefers going to those areas that animals are detected before. Meanwhile, the UAV explores other areas as well to cover the entire network and detects changes in the animal positions. We first define the mathematical model underlying the animal monitoring problem in terms of the value of information (VoI) and rewards. We propose a network model including clusters of sensor nodes and a single UAV that acts as a mobile sink and visits the clusters. Then, one MDP-based path planning approach is designed to maximize the VoI while reducing message delays. The effectiveness of the proposed approach is evaluated using two real-world movement datasets of zebras and leopard. Simulation results show that our approach outperforms greedy, random heuristics and the path planning based on the traveling salesman problem.

Rouhollah Rahmatizadeh, Pooya Abolghasemi, Aman Behal et al.

Robots assisting the disabled or elderly must perform complex manipulation tasks and must adapt to the home environment and preferences of their user. Learning from demonstration is a promising choice, that would allow the non-technical user to teach the robot different tasks. However, collecting demonstrations in the home environment of a disabled user is time consuming, disruptive to the comfort of the user, and presents safety challenges. It would be desirable to perform the demonstrations in a virtual environment. In this paper we describe a solution to the challenging problem of behavior transfer from virtual demonstration to a physical robot. The virtual demonstrations are used to train a deep neural network based controller, which is using a Long Short Term Memory (LSTM) recurrent neural network to generate trajectories. The training process uses a Mixture Density Network (MDN) to calculate an error signal suitable for the multimodal nature of demonstrations. The controller learned in the virtual environment is transferred to a physical robot (a Rethink Robotics Baxter). An off-the-shelf vision component is used to substitute for geometric knowledge available in the simulation and an inverse kinematics module is used to allow the Baxter to enact the trajectory. Our experimental studies validate the three contributions of the paper: (1) the controller learned from virtual demonstrations can be used to successfully perform the manipulation tasks on a physical robot, (2) the LSTM+MDN architectural choice outperforms other choices, such as the use of feedforward networks and mean-squared error based training signals and (3) allowing imperfect demonstrations in the training set also allows the controller to learn how to correct its manipulation mistakes.

Ladislau Bölöni

The Xapagy cognitive architecture had been designed to perform narrative reasoning: to model and mimic the activities performed by humans when witnessing, reading, recalling, narrating and talking about stories. Xapagy communicates with the outside world using Xapi, a simplified, "pidgin" language which is strongly tied to the internal representation model (instances, scenes and verb instances) and reasoning techniques (shadows and headless shadows). While not fully a semantic equivalent of natural language, Xapi can represent a wide range of complex stories. We illustrate the representation technique used in Xapi through examples taken from folk physics, folk psychology as well as some more unusual literary examples. We argue that while the Xapi model represents a conceptual shift from the English representation, the mapping is logical and consistent, and a trained knowledge engineer can translate between English and Xapi at near-native speed.

Saad Ahmad Khan, Ladislau Boloni

We describe an agent-based simulation of a fictional (but feasible) information trading business. The Gas Price Information Trader (GPIT) buys information about real-time gas prices in a metropolitan area from drivers and resells the information to drivers who need to refuel their vehicles. Our simulation uses real world geographic data, lifestyle-dependent driving patterns and vehicle models to create an agent-based model of the drivers. We use real world statistics of gas price fluctuation to create scenarios of temporal and spatial distribution of gas prices. The price of the information is determined on a case-by-case basis through a simple negotiation model. The trader and the customers are adapting their negotiation strategies based on their historical profits. We are interested in the general properties of the emerging information market: the amount of realizable profit and its distribution between the trader and customers, the business strategies necessary to keep the market operational (such as promotional deals), the price elasticity of demand and the impact of pricing strategies on the profit.