Because a diffusion model shares parameters for all diffusion steps, the noise schedule (parametrized by 1:T) is an important hyperparameter that determines how much weight we assign to each denoising problem. We find that standard noise schedules for continuous diffusions are not robust for text data. We hypothesize that the discrete nature of text and the rounding step make the model insensitive to noise near t =0 . Concretely, adding small amount of Gaussian noise to a word embedding is unlikely to change its nearest neighbor in the embedding space, making denoising an easy task near t =0 . To address this, we introduce a new sqrt noise schedule that is better suited for text, shown in Figure 5 defined by t =1 p t/T +s, where s is a small constant that corresponds to the starting noise level11. Compared to standard linear and cosine schedules, our sqrt schedule starts with a higher noise level and increase noise rapidly for the first 50 steps. Then sqrt slows down injecting noise to avoid spending much steps in the high-noise problems, which may be too difficult to solve well. The hyperparameters that are specific to Diffusion-LM include the number of diffusion steps, the architecture of the Diffusion-LM, the embedding dimension, and the noise schedule, . We set the diffusion steps to be 2000, the architecture to be BERT-base [7], and the sequence length to be 64. For the embedding dimensions, we select from d 2{ 16,64,128,256} and select d = 16for the E2E dataset and d = 128for ROCStories. For the noise schedule, we design the sqrt schedule (Appendix A) that is more robust to different parametrizations and embedding dimensions as shown in Appendix M. However, once we picked the x0-parametrization ( 4.2) the advantage of sqrt schedule is not salient. We train Diffusion-LMs using AdamW optimizer and a linearly decay learning rate starting at 1e-4, dropout of 0.1, batch size of 64, and the total number of training iteration is 200K for E2E dataset, and 800K for ROCStories dataset. Our Diffusion-LMs are trained on a single GPU: NVIDIARTXA5000, NVIDIAGeForce RTX 3090, or NVIDIAA100.

Improving the controllability, portability, and inference speed of diffusion language models (DLMs) is a key challenge in natural language generation. While recent research has shown significant success in complex text generation with language models, the memory and computational power are still very demanding and fall short of expectations, which naturally results in low portability and instability for the models. To mitigate these issues, numerous well-established methods were proposed for neural network quantization. To further enhance their portability of independent deployment as well as improve their stability evaluated by language perplexity, we propose a novel approach called the Quantized Embedding Controllable Diffusion Language Model (QE-CDLM). QE-CDLM builds upon the recent successful controllable DLMs by remodeling the task-specific embedding space via quantization. This leads to a gradient-based controller for the generation tasks, and more stable intermediate latent variables are obtained, which naturally brings in an accelerated convergence as well as better controllability. Additionally, the adaption fine-tuning method is employed to reduce tunable weights. Experimental results on five challenging fine-grained control tasks demonstrate that QE-CDLM compares favorably to existing methods in terms of quality and feasibility, achieving better perplexity and lightweight fine-tuning.

One of the main problems that local authorities of large cities have to face is the regulation of urban mobility. They need to provide the means to allow for the efficient movement of people and distribution of goods. However, the provisioning of transportation services needs to take into account general global objectives, like reducing emissions and having more healthy living environments, which may not always be aligned with individual interests. Urban mobility is usually provided through a transport infrastructure that includes all the elements that support mobility. On many occasions, the capacity of the elements of this infrastructure is lower than the actual demand and thus different transportation activities compete for their use. In this paper, we argue that scarce transport infrastructure elements should be assigned dynamically and in a prioritised manner to transport activities that have a higher utility from the point of view of society; for example, activities that produce less pollution and provide more value to society. In this paper, we define a general model for prioritizing the use of a particular type of transportation infrastructure element called time-unlimited elements, whose usage time is unknown a priori, and illustrate its dynamics through two use cases: vehicle-specific dynamic access restriction in city centres (i) based on the usage levels of available parking spaces and (ii) to assure sustained admissible air quality levels in the city centre. We carry out several experiments using the SUMO traffic simulation tool to evaluate our proposal.

Police will use live facial recognition technology in Cardiff during the Beyoncé concert on Wednesday, despite concerns about racial bias and human rights. The technology will be used in Cardiff city centre, but not at the stadium, to "support" the artist's concert at the Principality stadium by identifying wanted individuals and ensuring safeguarding, South Wales police said, as the artist kicks off the UK leg of her first solo headline tour in seven years. A spokesperson for the force said the technology would be used in the city centre, not at the concert itself. In the past, police use of live facial recognition (LFR) in England and Wales had been limited to special operations such as football matches or the coronation, when there was a crackdown on protesters. Daragh Murray, a senior lecturer of law at Queen Mary University in London, said the normalisation of invasive surveillance capability at events such as a concert was concerning, and was taking place without any real public debate.

Diffusion models have been successfully adapted to text generation tasks by mapping the discrete text into the continuous space. However, there exist nonnegligible gaps between training and inference, owing to the absence of the forward process during inference. Thus, the model only predicts based on the previously generated reverse noise rather than the noise computed by the forward process. Besides, the widely-used downsampling strategy in speeding up the inference will cause the mismatch of diffusion trajectories between training and inference. To understand and mitigate the above two types of training-inference discrepancies, we launch a thorough preliminary study. Based on our observations, we propose two simple yet effective methods to bridge the gaps mentioned above, named Distance Penalty and Adaptive Decay Sampling. Extensive experiments on \textbf{6} generation tasks confirm the superiority of our methods, which can achieve $100\times \rightarrow 200\times$ speedup with better performance.

Although the well-known MR-to-text E2E dataset has been used by many researchers, its MR-text pairs include many deletion/insertion/substitution errors. Since such errors affect the quality of MR-to-text systems, they must be fixed as much as possible. Therefore, we developed a refined dataset and some python programs that convert the original E2E dataset into a refined dataset.

Streets have been eerily quiet in recent months as coronavirus lockdowns imposed by governments around the world hit the pause button on normal life. And while many people have missed the shops and cafes, many have also appreciated the temporary respite from noise, pollution and congestion. As cities start to wake up from the so-called anthropause, questions are being being asked about how we can improve them more permanently. And the assumptions we had about making our cities smart may also need a rethink. Robots and drones have certainly come into their own during the global lockdown.

This paper presents an integrated methodology supporting decisions in urban planning. In particular, it deals with the prioritization and the selection of a portfolio of projects related to buildings of some values for the cultural heritage in cities. More precisely, our methodology has been validated to the historical center of Naples, Italy. Each project is assessed on the basis of a set of both quantitative and qualitative criteria with the purpose to determine their level of priority for further selection. This step was performed through the application of the Electre Tri-nC method which is a multiple criteria outranking based method for ordinal classification (or sorting) problems and allows to assign a priority level to each project as an analytical "recommendation" tool. To identify the efficient portfolios and to support the selection of the most adequate set of projects to activate, a set of resources (namely budgetary constraints) as well as some logical constraints related to urban policy requirements have to be taken into consideration together with the priority of projects in a portfolio analysis model. The process has been conducted by means of the interaction between analysts, municipality representative and experts. The proposed methodology is generic enough to be applied to other territorial or urban planning problems. We strongly believe that, given the increasing interest of historical cities to restore their cultural heritage, the integrated multiple criteria decision aiding analytical tool proposed in this paper has significant potential to be used in the future.

Driverless cars will cruise around city centres while their owners shop instead of parking to avoid extortionate parking fees, according to new research. The influx of the controversial vehicles meandering around the roads would likely increase congestion and cause huge traffic jams, scientists claim. During this time they will also save petrol by going slow to'kill time' - amplifying the issue. Driverless cars will cruise around city centres while their owners shop instead of parking to avoid extortionate parking fees, according to new research. Adam Millard-Ball, an associate professor of environmental studies at the University of California, Santa Cruz, said: 'Parking prices are what get people out of their cars and on to public transit.